Centaur robot quadruped and dual-arm force-position collaborative control method, device, equipment, medium, robot and system

Abstract

The application provides a method and device for force-position cooperative control of a centaur robot four-legged and double-armed, electronic equipment, a computer readable storage medium, a centaur robot and a system. The method obtains a four-legged support phase parameter and a double-armed end operation parameter of the centaur robot, and maps them into a closed loop kinematic chain; an action force transmission relationship atlas is constructed according to a path characteristic parameter; an initial stiffness parameter is configured according to a single leg swing mark or a full support mark, and a target force-position impedance adjustment parameter is generated based on the action force transmission relationship atlas, and then a posture compensation instruction and a gait stability maintaining instruction are generated, so that the double-armed and four-legged execute force-position cooperative operation.

Description

Technical Field

[0001] This application relates to the field of robot motion control technology, and more specifically, to a method, device, electronic device, computer-readable storage medium, centaur robot and system for force-position coordination control of the quadrupeds and arms of a centaur robot. Background Technology

[0002] With the development of legged mobile platforms, humanoid upper limb mechanisms, and high-degree-of-freedom actuators, hybrid robots that combine stable mobility and end-effector capabilities are increasingly being applied to technical scenarios such as handling in complex terrains, on-site assembly assistance, disaster environment operations, and unstructured scene operations. When performing tasks, these robots require both a lower support mechanism to maintain body posture and ground stability, and an upper operating mechanism to perform contact operations such as pushing, pulling, lifting, pressing, or grasping. Therefore, there is a strong dynamic coupling between the body support state and the end-effector operation state.

[0003] In existing composite robot control schemes, a separate approach is typically adopted, where upper-level manipulation control and lower-level motion control are handled separately. This scheme generally involves the upper-level manipulation controller first generating joint control parameters for the manipulator based on the end-effector's trajectory; then, the lower-level motion controller generates joint control parameters for the motion mechanism based on foot contact status or a preset gait pattern; finally, the overall robot controller coordinates and distributes these two types of control parameters according to preset priorities or fixed thresholds.

[0004] However, the aforementioned separate control scheme has significant technical drawbacks. When the upper operating mechanism performs contact actions such as pushing, pulling, and lifting, external interactive loads can affect the grounding state of the lower support mechanism through the body structure, easily causing gait fluctuations or foot slippage. Conversely, the body disturbances generated by the lower movement mechanism during movement or posture adjustment can also affect the end-effector positioning stability of the upper operating mechanism. Because existing schemes typically handle operation control and movement control separately, it is difficult to correlate external interactive loads, support states, and end-effector impedance adjustment processes within the same control link, resulting in low control adaptability of the composite robot in force-position cooperative operation scenarios. Summary of the Invention

[0005] This application provides a method, apparatus, electronic device, computer-readable storage medium, centaur robot and system for force-position coordination control of quadrupeds and arms, so as to at least alleviate the above-mentioned technical problems.

[0006] A method for force-position coordinated control of the quadrupeds and arms of a centaur robot, comprising: Obtain the quadrupedal support phase parameters and the end-effector operation parameters of the centaur robot, and map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. Based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs, a reaction force transmission relationship diagram characterizing the force distribution characteristics of the whole machine is constructed. The single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter is analyzed to identify the quadrupedal support state of the centaur robot, and the initial stiffness parameter is configured for the end of the two arms based on the single-leg swing mark or full-support mark. The initial stiffness parameter is compensated based on the reaction force transmission relationship diagram to generate the target force potential impedance adjustment parameter for the ends of the double arms. According to the target force-position impedance adjustment parameters, posture compensation commands for driving the joints of both arms are generated, and after the posture compensation commands are issued, gait stabilization commands for driving the joints of the four legs are generated, so that the arms and the four legs perform force-position coordinated operations.

[0007] Optionally, mapping the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain includes: Obtain the trunk centroid reference system and use the trunk centroid reference system as a unified spatial reference for the quadrupedal support phase parameter and the end-effector operation parameter of the two arms. Based on the operational parameters of the two arms, a positive kinematic link is established from the ends of the two arms to the reference system of the trunk center of mass; Based on the four-legged support phase parameters, an inverse kinematic link is established from each foot end of the four legs to the trunk center of mass reference system; Using the trunk centroid reference system as the splicing reference, the forward kinematic links and the reverse kinematic links are spliced ​​together to generate a global spatial topology structure that runs through the arms, trunk and four legs. Extract the ground contact force characteristics of each leg included in the four-legged support phase parameters; Extract the joint torque characteristics included in the end-effector operation parameters of the two arms; The ground contact force characteristics of each leg and the joint torque characteristics are configured into the corresponding nodes of the global spatial topology to generate the closed-loop kinematic chain.

[0008] Optionally, the step of constructing a reaction force transmission relationship map characterizing the force distribution characteristics of the entire machine based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs includes: Read the Jacobian transfer matrix elements in the closed-loop kinematic chain; Based on the Jacobian transfer matrix elements, the six-degree-of-freedom trunk perturbation vector caused by the interaction force at the ends of the arms is identified. Along the force transmission path of each leg joint of the four legs, the attenuation feature of the six-degree-of-freedom perturbation vector of the trunk is extracted to determine the attenuation coefficient of the six-degree-of-freedom perturbation vector of the trunk transmitted along each leg joint of the four legs. The attenuation coefficient is labeled as the path feature parameter; Using the six-degree-of-freedom perturbation vector of the torso and the path feature parameters as the basis for map construction, a mapping association table is constructed to record the additional perturbation components experienced by each of the four legs. Then, the reaction force transmission relationship map is generated based on the mapping association table.

[0009] Optionally, the step of parsing the single-leg swing marker or full-support marker carried in the quadrupedal support phase parameters to identify the quadrupedal support state of the centaur robot, and configuring initial stiffness parameters for the ends of the twin arms based on the single-leg swing marker or full-support marker, includes: The single-leg swing marker carried in the quadrupedal support phase parameter is analyzed to detect whether the quadrupedal support state of the centaur robot corresponds to the single-leg swing state. When the single-leg swing marker is detected, the centaur robot is determined to be in a walking state or a posture adjustment state based on the single-leg swing marker. Extract the first impedance reference profile stored in the first data pool, the first impedance reference profile containing the set compliance follow stiffness value; Using the set compliant following stiffness value as the stiffness source corresponding to the single-leg swing mark, the set compliant following stiffness value is assigned to the ends of both arms to generate the initial stiffness parameter corresponding to the single-leg swing mark, thereby enabling the ends of both arms to have compliant characteristics that absorb external vibrations.

[0010] Optionally, the step of parsing the single-leg swing marker or full-support marker carried in the quadrupedal support phase parameters to identify the quadrupedal support state of the centaur robot, and configuring initial stiffness parameters for the ends of the twin arms based on the single-leg swing marker or full-support marker, further includes: The full support marker carried in the quadrupedal support phase parameter is analyzed to detect whether the quadrupedal support state of the centaur robot corresponds to the full support state. When the full support marker is detected, it is determined that the Centaur robot is in a stationary working state based on the full support marker; Extract the second impedance reference profile stored in the second data pool. The second impedance reference profile contains positioning maintenance stiffness values ​​configured for positioning operations. The positioning maintenance stiffness values ​​have stronger disturbance rejection properties compared to the compliant following stiffness values. Using the positioning and maintaining stiffness value as the stiffness source corresponding to the full support mark, the positioning and maintaining stiffness value is assigned to the end of the dual arms to generate the initial stiffness parameter corresponding to the full support mark, thereby controlling the positioning accuracy of the end of the dual arms when performing operations.

[0011] Optionally, the step of compensating the initial stiffness parameter based on the reaction force transmission relationship diagram to generate target potential impedance adjustment parameters for the ends of the double arms includes: Extract the overall machine stiffness coupling matrix under the current attitude from the reaction force transmission relationship graph; Locate the stiffness element corresponding to the supporting foot in the overall machine stiffness coupling matrix; The stiffness element of the support foot is compared with a set standing threshold to determine whether the stiffness element of the support foot is lower than the set standing threshold. If the stiffness element of the support foot is lower than the set standing threshold, then the deviation characteristic value of the set standing threshold relative to the stiffness element of the support foot is calculated. A stiffness attenuation compensation amount is generated based on the deviation characteristic value; The initial stiffness parameter is reduced in order by using the stiffness attenuation compensation amount to generate the corrected impedance parameter, and the corrected impedance parameter is determined as the target force potential impedance adjustment parameter for the ends of the two arms.

[0012] Optionally, the step of generating posture compensation commands for driving the biarm joints according to the target force potential impedance adjustment parameters, and generating gait stabilization commands for driving the quadrupedal joints after the posture compensation commands are issued, includes: Based on the end stiffness coefficient and end damping coefficient in the target force-position impedance adjustment parameters, determine the dynamic offset of the desired end contact force relative to the actual end contact force. The dynamic offset is converted into the expected joint torque value of the two arm joints; The posture compensation command is generated based on the expected value of the joint torque, and the posture compensation command is sent to the hardware controller of the two arm joints; After the attitude compensation command is issued, the target force impedance adjustment parameter is projected onto the joint space to determine the coupling reaction torque; The additional support force command for the four feet is generated based on the coupled reaction torque, and the gait stabilization command is generated based on the additional support force command.

[0013] Optionally, the step of generating additional support force commands at the ends of the four feet based on the coupled reaction torque, and generating the gait stabilization commands based on the additional support force commands, includes: Establish the overall machine reaction torque balance constraint conditions; Within the overall reaction torque balance constraint, the coupled reaction torque is distributed to each foot end in contact with the ground to generate additional support force commands for the four feet. Obtain the desired ground reaction force for each of the said feet corresponding to the additional support force command; Verify that the desired ground reaction force is within the anti-slip friction constraint boundary; If the desired ground reaction force is within the anti-slip friction constraint boundary, then a torque drive signal for each of the four legs is generated based on the desired ground reaction force. The gait stabilization command is generated based on the torque drive signal and sent to the hardware controller of the four-legged joints to prevent slippage of the four legs.

[0014] Optionally, obtaining the quadrupedal support phase parameters and the end-effector manipulation parameters of the centaur robot includes: Contact force sensors deployed at the ends of the four feet collect the plantar contact pressure values; The foot end where the sole contact pressure value continuously exceeds the contact threshold is marked as being in the support phase; The foot end where the sole contact pressure value does not exceed the contact threshold is marked as being in the swing phase; The quadrupedal support phase parameters are constructed based on the support phase markers and swing phase markers corresponding to each foot end; Three-dimensional force data are collected by a six-axis torque sensor deployed at the ends of the twin arms; The joint rotation angle is acquired using a dual-arm joint encoder; The three-dimensional spatial force data and the joint rotation angle are integrated and processed to generate the end-effector operation parameters of the two arms.

[0015] A force-position coordination control device for the quadrupeds and arms of a centaur robot, comprising: The parameter acquisition module is configured to acquire the quadrupedal support phase parameters and the end-effector operation parameters of the Centaur robot. The closed-loop kinematic chain mapping module is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The relationship graph construction module is configured to construct a reaction force transmission relationship graph that characterizes the force distribution characteristics of the whole machine based on the path feature parameters that characterize the transmission of contact force at the ends of the two arms to the four legs in the closed-loop kinematic chain. The initial stiffness configuration module is configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms according to the single-leg swing mark or full-support mark. The impedance adjustment generation module is configured to compensate the initial stiffness parameter based on the reaction force transmission relationship spectrum to generate a target force potential impedance adjustment parameter for the ends of the double arms. The collaborative instruction generation module is configured to generate posture compensation instructions for driving the joints of both arms according to the target force-position impedance adjustment parameters, and to generate gait stabilization instructions for driving the joints of the four legs after the posture compensation instructions are issued, so that the arms and the four legs perform force-position collaborative operations.

[0016] An electronic device, including a memory and a processor; The memory stores computer programs; When the processor executes the computer program, it implements the above-mentioned force-position coordination control method for the quadrupeds and arms of the centaur robot.

[0017] A computer-readable storage medium storing a computer program; When the computer program is executed by the processor, it implements the aforementioned force-position coordination control method for the quadrupeds and arms of the centaur robot.

[0018] A centaur robot includes a robot torso, a quadruped mechanism, a dual-arm mechanism, a sensor acquisition component, a dual-arm joint hardware controller, a quadruped joint hardware controller, and a control processor. The quadruped mechanism is connected to the robot's torso, and the quadruped mechanism includes four leg joints and four leg ends; The dual-arm mechanism is connected to the robot's torso, and the dual-arm mechanism includes dual-arm joints and dual-arm ends; The sensing and acquisition component is configured to acquire the four-leg support phase parameters corresponding to the four-legged mechanism and the end-effector operation parameters corresponding to the dual-arm mechanism. The control processor is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The control processor is also configured to construct a reaction force transmission relationship map representing the force distribution characteristics of the whole machine based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs. The control processor is also configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms according to the single-leg swing mark or full-support mark. The control processor is also configured to compensate the initial stiffness parameter based on the reaction force transmission relationship diagram to generate a target force potential impedance adjustment parameter for the ends of the two arms. The control processor is also configured to generate attitude compensation commands for driving the bi-arm joints according to the target force impedance adjustment parameters, and to send the attitude compensation commands to the bi-arm joint hardware controller. The control processor is also configured to generate gait stabilization commands for driving the quadrupedal joints after the attitude compensation command is issued, and to issue the gait stabilization commands to the quadrupedal joint hardware controller so that the dual-arm mechanism and the quadrupedal mechanism perform force-position coordinated operations.

[0019] A force-position coordination control system for the quadrupedal and dual arms of a centaur robot includes a centaur robot, a parameter acquisition unit, a coordination control unit, a dual-arm joint hardware controller, and a quadrupedal joint hardware controller. The centaur robot includes a quadruped mechanism and a dual-arm mechanism; The parameter acquisition unit is configured to acquire the four-legged support phase parameters corresponding to the four-legged mechanism and the double-arm end operation parameters corresponding to the double-arm mechanism. The collaborative control unit is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The collaborative control unit is also configured to construct a reaction force transmission relationship map representing the force distribution characteristics of the whole machine based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs. The collaborative control unit is also configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms of the dual-arm mechanism according to the single-leg swing mark or the full-support mark. The collaborative control unit is also configured to compensate the initial stiffness parameter based on the reaction force transmission relationship diagram to generate a target force potential impedance adjustment parameter for the ends of the two arms. The collaborative control unit is also configured to generate attitude compensation commands for driving the dual-arm joints of the dual-arm mechanism according to the target force impedance adjustment parameters, and send the attitude compensation commands to the dual-arm joint hardware controller. The collaborative control unit is also configured to generate gait stabilization commands for driving the quadrupedal joints of the quadrupedal mechanism after the attitude compensation command is issued, and send the gait stabilization commands to the quadrupedal joint hardware controller so that the dual-arm mechanism and the quadrupedal mechanism perform force-position collaborative operations.

[0020] The technical advantages of the technical solution provided in this application are: The force-position coordination control method for the quadrupedal and dual-arm centaur robot provided in this application addresses the technical deficiency of existing separate control schemes, which struggle to correlate external interactive loads, support states, and end-effector impedance adjustment processes within the same control link. By acquiring the quadrupedal support phase parameters and dual-arm end-effector operation parameters of the centaur robot, and mapping these parameters into a closed-loop kinematic chain, the quadrupedal support state and dual-arm end-effector operation state are no longer processed in isolation. Compared to traditional schemes where upper operation control and lower movement control generate control quantities separately, this application first incorporates the quadrupedal support phase parameters and dual-arm end-effector operation parameters into the closed-loop kinematic chain. This allows subsequent control processing to be simultaneously based on the motion correlation between the quadrupedal and dual-arms, improving the coordination foundation between quadrupedal support control and dual-arm end-effector control from a technical processing entry point.

[0021] Furthermore, this application constructs a reaction force transmission relationship map characterizing the force distribution characteristics of the entire machine based on the path characteristic parameters representing the transmission of contact force from the ends of the two arms to the four legs in the closed-loop kinematic chain. Therefore, when contact operations such as pushing, pulling, and lifting are performed at the ends of the two arms and external interactive loads are generated, these external interactive loads are no longer considered merely disturbances in the local control of the two arms, but are associated with the force distribution characteristics of the entire machine through the path characteristic parameters. Compared to traditional solutions that lack an expression of external load transmission relationships, the reaction force transmission relationship map can provide a force distribution basis for subsequent compensation of initial stiffness parameters, thereby alleviating the problem of gait fluctuations or foot slippage easily caused when external interactive loads affect the support state of the four legs through the fuselage structure.

[0022] Furthermore, this application analyzes the single-leg swing marker or full-support marker carried in the quadrupedal support phase parameters to identify the quadrupedal support state of the Centaur robot, and configures the initial stiffness parameters of the end arms based on the single-leg swing marker or the full-support marker. Thus, the stiffness configuration of the end arms no longer adopts a fixed configuration independent of the support state, but can form initial stiffness parameters based on the quadrupedal support state. Compared to traditional solutions where upper-level operation control struggles to perceive changes in lower-level movement or stance, this application establishes a correspondence between the initial stiffness parameters of the end arms and the single-leg swing marker or full-support marker, thereby mitigating the interference of quadrupedal movement or posture adjustment processes on the positioning stability of the end arms.

[0023] Furthermore, this application compensates the initial stiffness parameter based on the reaction force transmission relationship diagram to generate target force-potential impedance adjustment parameters for the ends of the twin arms. Thus, the initial stiffness parameter not only originates from the four-legged support state but is also further constrained by the overall machine's force distribution characteristics, enabling the force-potential impedance adjustment parameters at the ends of the twin arms to simultaneously reflect both the support state and the external interactive load transmission state. Compared to traditional methods that rely solely on the end-effector trajectory or fixed impedance parameters for operational control, this application can adaptively adjust the force-potential impedance at the ends of the twin arms when the contact force at the ends affects the four-legged support, resulting in a higher degree of coupling and adaptation between the twin-arm end-effector control and the four-legged support control.

[0024] Furthermore, this application generates posture compensation commands for driving the bi-arm joints according to the target force-position impedance adjustment parameters, and generates gait stabilization commands for driving the quadrupedal joints after the posture compensation commands are issued, so that the bi-arms and quadrupedal joints perform force-position coordinated operations. This processing sequence makes the posture compensation of the bi-arm joints and the gait stabilization of the quadrupedal joints no longer unrelated parallel control actions, but a coordinated control process that is sequentially connected around the target force-position impedance adjustment parameters. Compared with the traditional approach of coordinating the issuance of control quantities according to preset priorities or fixed thresholds, this application can further transmit the force-position adjustment results of the bi-arm end-effectors to the gait stabilization command generation process of the quadrupedal joints, thereby improving the matching degree between the quadrupedal support stability and the bi-arm end-effector operation stability when the composite robot performs force-position coordinated operations. Attached Figure Description

[0025] Figure 1 This application provides an embodiment of a force-position coordinated control scenario for a centaur robot's four legs and two arms. Figure 2 This application provides a method for force-position coordinated control of the quadruped and double arms of a centaur robot. Figure 3 This application provides an embodiment of a force-position coordination control device for the quadrupedal legs and dual arms of a centaur robot. Figure 4 An electronic device is described in an embodiment of this application; Figure 5 This application provides an embodiment of a computer-readable storage medium. Figure 6 This application provides an embodiment of a mobile robot. Figure 7 This application provides a force-position coordination control system for a centaur robot with four legs and two arms. Detailed Implementation

[0026] like Figure 1The image shows a force-position coordinated control scenario of the quadrupedal legs and dual arms of a centaur robot according to an embodiment of this application; as shown... Figure 2 As shown, this embodiment of the present application discloses a force-position coordination control method for the quadrupedal and dual-arm centaur robot, comprising: Obtain the quadrupedal support phase parameters and the end-effector operation parameters of the centaur robot, and map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. Based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs, a reaction force transmission relationship diagram characterizing the force distribution characteristics of the whole machine is constructed. The single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter is analyzed to identify the quadrupedal support state of the centaur robot, and the initial stiffness parameter is configured for the end of the two arms based on the single-leg swing mark or full-support mark. The initial stiffness parameter is compensated based on the reaction force transmission relationship diagram to generate the target force potential impedance adjustment parameter for the ends of the double arms. According to the target force-position impedance adjustment parameters, posture compensation commands for driving the joints of both arms are generated, and after the posture compensation commands are issued, gait stabilization commands for driving the joints of the four legs are generated, so that the arms and the four legs perform force-position coordinated operations.

[0027] Optionally, mapping the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain includes: Obtain the trunk centroid reference system and use the trunk centroid reference system as a unified spatial reference for the quadrupedal support phase parameter and the end-effector operation parameter of the two arms. Based on the operational parameters of the two arms, a positive kinematic link is established from the ends of the two arms to the reference system of the trunk center of mass; Based on the four-legged support phase parameters, an inverse kinematic link is established from each foot end of the four legs to the trunk center of mass reference system; Using the trunk centroid reference system as the splicing reference, the forward kinematic links and the reverse kinematic links are spliced ​​together to generate a global spatial topology structure that runs through the arms, trunk and four legs. Extract the ground contact force characteristics of each leg included in the four-legged support phase parameters; Extract the joint torque characteristics included in the end-effector operation parameters of the two arms; The ground contact force characteristics of each leg and the joint torque characteristics are configured into the corresponding nodes of the global spatial topology to generate the closed-loop kinematic chain.

[0028] Preferably, the specific implementation process of the step "obtaining the torso center of mass reference system and using the torso center of mass reference system as a unified spatial reference for the quadrupedal support phase parameter and the dual-arm end-effector operation parameter" is as follows: After the Centaur robot enters the current control cycle, the mass distribution calibration record formed during the assembly calibration stage and the torso posture sampling record formed during the current control cycle are read first. The mass distribution calibration record includes the component mass distribution, component installation position, and joint connection sequence of the robot torso, quadrupedal mechanism, and dual-arm mechanism during the assembly calibration stage. The component mass distribution, component installation position, and joint connection sequence are used to jointly determine the static spatial assembly relationship between the robot torso, quadrupedal mechanism, and dual-arm mechanism. The torso posture sampling record includes the robot torso's posture angle, posture angle change rate, and torso inertial measurement data during the current control cycle. The posture angle, posture angle change rate, and torso inertial measurement data are used to jointly determine the real-time posture relationship of the robot torso during the current control cycle. The mass distribution calibration record is not a static file used solely to describe the robot structure, but rather to determine the spatial reference source for the robot torso, quadruped mechanism, and dual-arm mechanism under the same mass distribution caliber. Similarly, the torso posture sampling record is not data used solely for posture display, but rather to convert the static spatial assembly relationship into a real-time posture relationship under the current control cycle. Subsequently, the mass distribution calibration record and the torso posture sampling record are time-corresponded to form a spatial reference generation record. This spatial reference generation record includes the current control cycle marker, the torso posture sampling record, the mass distribution calibration record, and the center of mass position of the robot torso determined jointly by the mass distribution calibration record and the torso posture sampling record. When determining the torso center of mass reference system based on the spatial reference generation record, the center of mass position of the robot torso is first used as the spatial origin of the torso center of mass reference system. Then, the forward, lateral, and vertical directions of the robot torso are used to determine the spatial reference directions of the torso center of mass reference system, enabling the torso center of mass reference system to express the spatial position and posture direction of the robot torso under the current control cycle. Since the quadrupedal support phase parameter originates from the contact state of the quadrupedal feet, and the end-effector operation parameter originates from the force state of the end-effectors and the motion state of the end-effector joints, if the quadrupedal support phase parameter and the end-effector operation parameter are processed separately along their respective local coordinates, there will be a lack of a common spatial reference between the external interactive load on the end-effectors and the grounding relationship of the quadrupedal feet. Therefore, this application uses the torso centroid reference system as a unified spatial reference for the quadrupedal support phase parameter and the end-effector operation parameter, so that the positional relationship, directional relationship, and force transmission direction relationship between the quadrupedal feet, the end-effectors, and the robot torso can all be expressed under the same spatial reference.The technical essence of the unified spatial reference is to convert data from different sensor sources, different installation locations, and different kinematic links into the trunk centroid reference system, so that the subsequently established forward kinematic links and reverse kinematic links can be spliced ​​around the same trunk centroid reference system, thereby providing a common spatial basis for generating a global spatial topology structure that runs through the arms, trunk, and four legs.

[0029] Preferably, in the specific technical implementation of the step "establishing a positive kinematic link from the end-effectors of both arms to the trunk center of mass reference system based on the end-effector operation parameters", the three-dimensional spatial force data, joint rotation angles, and end-effector contact state fields in the end-effector operation parameters are first read, and the three-dimensional spatial force data, joint rotation angles, and end-effector contact state fields are written into the end-effector kinematic chain generation record. The end-effector kinematic chain generation record is used to include the external interactive loads on the end-effectors of both arms during the current control cycle, the joint rotation angles of each joint of both arms, and whether the end-effectors are in a contact operation state into the same positive kinematic link generation caliber, avoiding the omission of the influence of external interactive loads on the end-effectors on subsequent force-position coordination operations when establishing a positive kinematic link based solely on joint rotation angles. Subsequently, the dual-arm joint connection record generated during the assembly and calibration phase is read. This record includes the connection sequence, rotation axis direction, and installation distance between adjacent joints of each arm relative to the robot torso. The connection record is then correlated with the dual-arm kinematic chain generation record to determine the spatial transmission sequence of the dual-arm mechanism from the robot torso-side connection end to the arm ends. Based on this spatial transmission sequence, the joint rotation angle of each arm joint is sequentially converted into a corresponding arm joint posture transformation relationship. These posture transformation relationships are then sequentially connected according to the connection sequence in the connection record to form a dual-arm joint spatial transmission chain. The technical essence of this spatial transmission chain is a joint-by-joint description of the spatial position, spatial direction, and force direction of the arm ends relative to the robot torso, rather than simply recording a single position point of the arm ends. Subsequently, the torso-side endpoint of the spatial transmission chain is mapped to the torso center of mass reference system to form a positive kinematic link from the arm ends to the torso center of mass reference system. Each link node in the forward kinematics link corresponds to a biarm joint or biarm end effector, and the link segment between two adjacent link nodes corresponds to the biarm joint posture transformation relationship between adjacent structures. The three-dimensional spatial force data is configured into the link nodes corresponding to the biarm end effectors, and the joint rotation angle is configured into the link nodes corresponding to each biarm joint. Thus, the forward kinematics link not only reflects the positional relationship of the biarm end effectors in the trunk centroid reference system, but also retains the source of external interactive loads of the biarm end effectors in the biarm end effector operation parameters, providing a continuous spatial transmission basis for subsequently configuring the joint torque characteristics into the corresponding nodes of the global spatial topology.

[0030] Preferably, the specific implementation process of the step "establishing the inverse kinematic link from each foot end of the four legs to the trunk center of mass reference system based on the four-legged support phase parameters" is as follows: After the four-legged support phase parameters have been constructed by the support phase markers and swing phase markers corresponding to each foot end, the foot end support state field, plantar contact pressure value, and current control cycle marker in the four-legged support phase parameters are read first, and the foot end support state field, plantar contact pressure value, and current control cycle marker are written into the four-legged kinematic chain generation record. The foot end support state field is used to distinguish whether each foot end is in the support phase or swing phase within the current control cycle, the plantar contact pressure value is used to express the contact intensity between each foot end and the ground, and the current control cycle marker is used to ensure that the four-legged kinematic chain generation record and the trunk center of mass reference system are at the same time caliber. Subsequently, the quadrupedal joint connection records generated during the assembly and calibration phase are read. These records include the joint connection sequence, rotation direction, and installation distance between adjacent leg joints for each leg end corresponding to the quadrupedal foot end. Based on these records, a foot-end spatial backtracking chain is established for each quadrupedal foot end, pointing from the foot end to the robot's torso. The leg joint posture transformation relationships are then written into these backtracking chains according to the joint connection sequence in the quadrupedal joint connection records. The reason for establishing the foot-end spatial backtracking chain from the foot end to the robot's torso is that the quadrupedal support phase parameters primarily reflect the contact state between the foot end and the ground. Only by backtracking from the foot end in contact with the ground can the plantar contact pressure value, support phase marker, and swing phase marker be directly incorporated into the generation process of the inverse kinematics link. Subsequently, the robot torso-side connection end of each foot-end spatial backtracking chain is spatially mapped to the torso center of mass reference system to form a reverse kinematic link from each foot end to the torso center of mass reference system. The technical essence of this reverse kinematic link is to transmit the contact state and contact intensity of each foot end within the current control cycle back to the torso center of mass reference system along the leg joint structure, allowing the supporting effect of the foot ends on the robot torso to be expressed under a unified spatial reference. Since the subsequent generation of closed-loop kinematic chains requires simultaneously expressing the action paths of the arm ends on the robot torso and the supporting paths of the foot ends on the robot torso, the reverse kinematic link can form a spatial connection with the forward kinematic link at the torso center of mass reference system. This allows the four-leg support phase parameter to no longer exist merely as a gait state identifier, but to participate in the force-position coordination data construction process between the arms and four legs.

[0031] Preferably, the specific implementation process of the step "using the trunk centroid reference system as the splicing reference, splicing the forward kinematic links and the reverse kinematic links to generate a global spatial topology structure that runs through the arms, trunk, and four legs" is as follows: After the forward kinematic links and the reverse kinematic links have been formed respectively, the trunk-side endpoints, arm-end link nodes, and each arm joint link node of the forward kinematic links, as well as the trunk-side endpoints, each four-leg foot-end link node, and each leg joint link node of the reverse kinematic links are read, and the trunk-side endpoints, arm-end link nodes, each arm joint link node, trunk-side endpoints, each four-leg foot-end link node, and each leg joint link node of the reverse kinematic links are written into the spatial topology splicing record. The spatial topology stitching record is used to incorporate the forward kinematic links from the bi-arm mechanism and the reverse kinematic links from the quadruped mechanism into the same stitching interface. The trunk centroid reference system is used as a common stitching node in the spatial topology stitching record, rather than as an additional independent spatial point. Subsequently, the coordinate correspondence between the trunk-side endpoints of the forward kinematic links and the trunk centroid reference system is verified, as well as the coordinate correspondence between the trunk-side endpoints of the reverse kinematic links and the trunk centroid reference system. When both coordinate correspondences belong to the same time caliber defined by the current control cycle marker, the trunk-side endpoints of the forward kinematic links are connected to the trunk centroid reference system, and the trunk-side endpoints of the reverse kinematic links are also connected to the trunk centroid reference system to form the link stitching result. The connection relationships in the link splicing result express the spatial transmission order between the end arms, arm joints, robot torso, leg joints, and quadruped feet. The link splicing result does not change the joint connection order of the forward and reverse kinematic links, but rather establishes a common splicing node at the torso center of mass reference system. Subsequently, the link splicing result undergoes node organization processing, organizing the end arms, each arm joint, torso center of mass reference system, each leg joint, and each quadruped foot into topological nodes in the global spatial topology structure. The spatial pose transformation relationships between adjacent topological nodes are then organized into topological connection relationships in the global spatial topology structure, thereby generating a global spatial topology structure that runs through the arms, torso, and quadrupeds. The technical essence of the global spatial topology is a data structure that places the forward kinematic links and the reverse kinematic links in the same spatial organization, with the torso center of mass reference system as the center. In this global spatial topology, the ground contact force features of each leg extracted subsequently can be configured into the topology nodes corresponding to the feet of the four legs, and the joint torque features extracted subsequently can be configured into the topology nodes corresponding to the joints of the two arms, thereby enabling the closed-loop kinematic chain to simultaneously express the spatial connection relationship and the force source relationship.

[0032] Preferably, the specific implementation process of the step "extracting the ground contact force characteristics of each leg included in the four-leg support phase parameters" is as follows: After the four-leg support phase parameters have been incorporated into the spatial organization relationship corresponding to the global spatial topology, the plantar contact pressure value, support period marker, swing period marker, and four-leg foot end identifier in the four-leg support phase parameters are read, and the plantar contact pressure value, support period marker, swing period marker, and four-leg foot end identifier are written into the ground contact force extraction record. The ground contact force extraction record is used to correspond the source of the plantar contact pressure value, the source of the support state, and the topological node position in the global spatial topology for each four-leg foot end, so that the ground contact force characteristics of each leg can be subsequently configured into the corresponding node of the global spatial topology, rather than remaining at the level of sensor acquisition data. Subsequently, the ground contact force extraction records undergo effective contact screening. This screening uses the contact threshold used when constructing the quadrupedal support phase parameters as the screening criterion. Quadrupedal feet in the support phase with plantar contact pressure values ​​meeting the contact threshold condition are identified as effective support feet. For quadrupedal feet in the swing phase, their plantar contact pressure values ​​are not considered as effective support sources in the formation of each leg's ground contact force characteristics, but the corresponding swing phase marker is still retained in the ground contact force extraction records to distinguish between effective support feet and quadrupedal feet in the swing phase in the global spatial topology. Next, the plantar contact pressure values ​​corresponding to the effective support feet undergo direction assignment processing. The plantar contact pressure values ​​along the vertical direction of the torso centroid reference system are determined as the main ground support component. Combined with the spatial direction relationship of the quadrupedal feet in the reverse kinematic link, the ground contact force direction relationship of the corresponding leg is formed. Based on the main ground support component, the ground contact force direction relationship, and the quadrupedal foot markers, the ground contact force characteristics of each leg are formed. The technical essence of the ground contact force characteristics of each leg is to convert the support state and foot contact pressure values ​​in the quadrupedal support phase parameters into nodal force descriptions that can be configured into the global spatial topology. This gives each leg a clear data source and spatial attribution for its support contribution to the robot's torso within the current control cycle. Through these leg ground contact force characteristics, the subsequently generated closed-loop kinematic chain can distinguish which quadrupedal foot is providing ground support, which quadrupedal foot is in a swinging phase, and the directional relationship of the contact force of the effective supporting foot along the torso's center of mass reference system. This provides a force basis directly corresponding to the quadrupedal support state for subsequently constructing a reaction force transmission relationship diagram based on the closed-loop kinematic chain.

[0033] Preferably, the specific implementation process of step "extracting the joint torque features contained in the end-effector manipulation parameters" is as follows: After the end-effector manipulation parameters have participated in establishing the positive kinematic link, the three-dimensional spatial force data, joint rotation angle, end-effector contact state field, and end-effector joint identifier in the end-effector manipulation parameters are read, and the three-dimensional spatial force data, joint rotation angle, end-effector contact state field, and end-effector joint identifier are written into the joint torque extraction record. The joint torque extraction record is used to correlate the external interactive loads on the end-effectors with the joint rotation angles of each end-effector joint in the current control cycle, so that the joint torque features can be traced back to the end-effector manipulation parameters along the positive kinematic link, rather than being generated in isolation by individual joint drive data. Subsequently, based on the bi-arm joint spatial transmission chain in the positive kinematic link, the three-dimensional spatial force data is subjected to joint spatial attribution processing. This joint spatial attribution processing does not directly equate the three-dimensional spatial force data with joint torque; instead, it maps the external interactive loads on the bi-arm ends to the corresponding bi-arm joint positions step-by-step according to the spatial directional relationship between the bi-arm ends and each bi-arm joint and the joint lever arm direction. Since the joint rotation angle reflects the spatial posture of the bi-arm joints within the current control cycle, the joint rotation angle participates in determining the joint lever arm direction corresponding to each bi-arm joint. Since the three-dimensional spatial force data reflects the magnitude and direction of the contact force on the bi-arm ends, the three-dimensional spatial force data participates in determining the source of joint torque that each bi-arm joint needs to counteract or follow. Subsequently, the joint lever arm direction corresponding to each bi-arm joint, the contact force direction of the three-dimensional spatial force data, and the bi-arm joint identifier are processed accordingly to form joint torque characteristics. The technical essence of these joint torque characteristics is to express the torque influence of the external interactive loads on each bi-arm joint under the trunk center-of-mass reference system, rather than simply expressing the driving state of the bi-arm joints themselves. By configuring the joint torque features to the corresponding bi-arm joint nodes in the global spatial topology, the subsequently generated closed-loop kinematic chain can simultaneously include the external interactive load sources of the bi-arm end-effectors during the bi-arm end-effector operation, the spatial attitude sources of the bi-arm joints, and the spatial transmission relationship between the bi-arm joints and the robot torso. In this way, the joint torque features, together with the ground contact force features of each leg, can participate in the formation of the closed-loop kinematic chain, establishing a correspondence between the contact operation state of the bi-arm end-effectors and the support state of the four legs in the same data structure.

[0034] Preferably, the specific implementation process of the step "configuring the ground contact force characteristics of each leg and the joint torque characteristics into the corresponding nodes of the global spatial topology to generate the closed-loop kinematic chain" is as follows: After the global spatial topology, the ground contact force characteristics of each leg, and the joint torque characteristics have been formed, the topology node type, topology node position, and topology connection relationship in the global spatial topology are read first, and the four-leg foot end identifier, main ground support component, and ground contact force direction relationship in the ground contact force characteristics of each leg are read simultaneously, as well as the double-arm joint identifier, joint lever arm direction, and the correspondence of external interactive loads at the ends of the double arms in the joint torque characteristics. Subsequently, based on the quadrupedal foot-end identifiers, the corresponding quadrupedal foot-end nodes are located in the global spatial topology, and the main grounding support components and the grounding contact force direction relationship are configured into the corresponding quadrupedal foot-end nodes to form quadrupedal foot-end nodes with grounding force attributes. Similarly, based on the bi-arm joint identifiers, the corresponding bi-arm joint nodes are located in the global spatial topology, and the joint lever arm direction and the correspondence of the external interactive loads at the bi-arm ends are configured into the corresponding bi-arm joint nodes to form bi-arm joint nodes with joint torque attributes. The determination of the corresponding nodes is not a simple match based on node names, but rather a matching based on topology node type, topology node position, and the current control cycle marker, ensuring that the grounding contact force characteristics and joint torque characteristics of each leg configured in the global spatial topology are consistent with the spatial relationships within the current control cycle. Subsequently, along the topological connection relationships in the global spatial topology, link connection processing is performed on the bi-arm joint nodes with joint torque attributes, the torso centroid reference system, and the quadrupedal foot-end nodes with grounding force attributes to generate the closed-loop kinematic chain. The technical essence of the closed-loop kinematic chain is to organize the joint torque characteristics corresponding to the end-effector operation parameters of the two arms, the spatial reference relationship corresponding to the robot torso, and the ground contact force characteristics of each leg corresponding to the quadrupedal support phase parameters into a traceable spatial force link under the same torso center of mass reference system. Since the closed-loop kinematic chain retains the forward kinematic link from the end-effectors to the robot torso, and also retains the reverse kinematic link from the quadrupedal feet to the robot torso, and further configures the ground contact force characteristics and joint torque characteristics of each leg into the corresponding nodes, the closed-loop kinematic chain can express the transmission path of the end-effector contact operation to the quadrupedal support state, and also express the source of the constraint of the quadrupedal support state on the attitude adjustment of the end-effectors.Subsequently, when constructing the reaction force transmission relationship map based on the path characteristic parameters representing the transmission of contact force from the ends of the two arms to the four legs in the closed-loop kinematic chain, the corresponding relationships between the forward kinematic link, the reverse kinematic link, the joint nodes of the two arms with joint torque attributes, and the foot end nodes of the four legs with grounding force attributes can be directly read from the closed-loop kinematic chain. This gives the reaction force transmission relationship map a data basis consistent with the force-position collaborative operation state within the current control cycle.

[0035] Optionally, the step of constructing a reaction force transmission relationship map characterizing the force distribution characteristics of the entire machine based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs includes: Read the Jacobian transfer matrix elements in the closed-loop kinematic chain; Based on the Jacobian transfer matrix elements, the six-degree-of-freedom trunk perturbation vector caused by the interaction force at the ends of the arms is identified. Along the force transmission path of each leg joint of the four legs, the attenuation feature of the six-degree-of-freedom perturbation vector of the trunk is extracted to determine the attenuation coefficient of the six-degree-of-freedom perturbation vector of the trunk transmitted along each leg joint of the four legs. The attenuation coefficient is labeled as the path feature parameter; Using the six-degree-of-freedom perturbation vector of the torso and the path feature parameters as the basis for map construction, a mapping association table is constructed to record the additional perturbation components experienced by each of the four legs. Then, the reaction force transmission relationship map is generated based on the mapping association table.

[0036] Preferably, the specific implementation process of the step "reading the Jacobian transfer matrix elements in the closed-loop kinematic chain" is as follows: After the closed-loop kinematic chain has been formed by the forward kinematic link, the reverse kinematic link, the bi-arm joint nodes with joint torque attributes, and the quadrupedal foot nodes with grounding force attributes, the topology node type, topology node position, topology connection relationship, bi-arm joint posture transformation relationship, leg joint posture transformation relationship, joint torque characteristics, and grounding contact force characteristics of each leg in the closed-loop kinematic chain are read first. The topology node type, topology node position, topology connection relationship, bi-arm joint posture transformation relationship, leg joint posture transformation relationship, joint torque characteristics, and grounding contact force characteristics of each leg are then written into the Jacobian transfer matrix generation record. The Jacobian transfer matrix generation record is used to incorporate the topological connections and force sources in the closed-loop kinematic chain into the same differential transfer caliber. This differential transfer caliber expresses the local correspondence between node position changes, node orientation changes, and node force changes, ensuring that the subsequently formed Jacobian transfer matrix elements are not isolated mathematical entries but can be traced back to the biarm joint nodes, trunk centroid reference system, leg joint nodes, and quadruped foot nodes in the closed-loop kinematic chain. Subsequently, using the trunk centroid reference system as a unified spatial reference, the biarm joint posture transformation relationship and the leg joint posture transformation relationship are processed using micro-displacement response expansion. The technical essence of this micro-displacement response expansion processing is that when a small position or orientation change occurs in a biarm joint node or leg joint node during the current control cycle, the trunk centroid reference system and the associated quadruped foot nodes generate corresponding spatial response relationships under the same spatial reference. Since the closed-loop kinematic chain has already configured the joint torque characteristics into the biarm joint nodes and the leg ground contact force characteristics into the quadrupedal foot nodes, when generating the Jacobian transfer matrix elements, it is also necessary to map the joint torque characteristics to the biarm joint posture transformation relationship and the leg ground contact force characteristics to the leg joint posture transformation relationship, so that each Jacobian transfer matrix element can express the position response, direction response, or force response when a node disturbance is transmitted to another node along a topological connection relationship. The technical essence of the Jacobian transfer matrix elements is the local transmission sensitivity data of the closed-loop kinematic chain in the current control cycle; wherein, the row where the Jacobian transfer matrix element is located corresponds to the spatial response component in the trunk centroid reference system, the leg joint node, or the quadrupedal foot node, the column where the Jacobian transfer matrix element is located corresponds to the disturbance source component introduced by the interaction force of the biarm joint node or the biarm end, and the Jacobian transfer matrix element at the intersection of the row and column is used to express the transmission contribution of the disturbance source component to the spatial response component.Based on the local transfer sensitivity data, the Jacobian transfer matrix elements can subsequently be directly used to confirm the six-degree-of-freedom perturbation vector of the torso and provide a source of calculation for the path characteristic parameters of the interaction force transmitted from the ends of the arms to the four legs.

[0037] Preferably, in the specific technical implementation of the step "confirming the six-degree-of-freedom perturbation vector of the torso caused by the interaction force on the ends of the two arms according to the Jacobian transfer matrix elements", three-dimensional spatial force data is first read from the operation parameters of the ends of the two arms, and the three-dimensional spatial force data is correspondingly processed with the nodes of the ends of the two arms in the closed-loop kinematic chain to form a dual-arm end interaction force reading record. The dual-arm end interaction force reading record is used to characterize the magnitude, direction, and location of the interaction force on the ends of the two arms in the current control cycle; the interaction force on the ends of the two arms is expressed by the three-dimensional spatial force data at the data level, and by the nodes of the ends of the two arms in the closed-loop kinematic chain at the spatial level. Therefore, the dual-arm end interaction force reading record can map the three-dimensional spatial force data and the nodes of the ends of the two arms to the same torso center of mass reference system. Subsequently, based on the positive kinematic link between the end-effectors of the two arms and the trunk center of mass reference system, the Jacobian transfer matrix element corresponding to the positive kinematic link is read from the Jacobian transfer matrix generation record, and the interaction force on the end-effectors of the two arms along the positive kinematic link is converted into the torque influence source under the trunk center of mass reference system. This conversion process does not directly mix the dimensions of force and displacement, but first determines the corresponding torque influence source based on the direction and position of the interaction force on the end-effectors of the two arms, and then maps the torque influence source to the translational and rotational perturbation components of the robot trunk based on the local transfer sensitivity data expressed by the Jacobian transfer matrix elements. The resulting six-DOF perturbation vector for the torso includes translational perturbation components along the forward, lateral, and vertical directions of the torso center-of-mass reference system, as well as rotational perturbation components around the torso center-of-mass reference system. Each translational and rotational perturbation component originates from the interaction forces experienced by the end-effectors and are transmitted to the torso center-of-mass reference system via the positive kinematic link. The technical essence of this six-DOF perturbation vector for the torso is to transform the interaction forces experienced by the end-effectors due to end-effector contact operations from the end-effector nodes into a spatial perturbation description at the robot's torso level. This allows subsequent analysis to move beyond judging the impact of end-effector contact operations solely from the perspective of the end-effector joints themselves, and instead to analyze the force transmission paths of each leg joint from the perspective of the robot's torso being disturbed. Because the interaction force at the ends of the arms of the centaur robot during contact operations such as pushing, pulling, and lifting will affect the grounding state of the four legs through the robot's torso, the six-degree-of-freedom perturbation vector of the torso can provide a unified perturbation entry point for the subsequent extraction of attenuation features along the force transmission path of each leg joint.

[0038] Preferably, the specific implementation process of the step "extracting attenuation features from the six-degree-of-freedom perturbation vector of the torso along the force transmission path of each leg joint of the four legs to determine the attenuation coefficient of the six-degree-of-freedom perturbation vector of the torso along the force transmission path of each leg joint of the four legs" is as follows: After the six-degree-of-freedom perturbation vector of the torso has been formed, the arrangement order of the leg joint nodes, the leg joint posture transformation relationship, the topological connection relationship, and the four-leg foot end nodes with grounding force attributes between the torso centroid reference system and each foot end node in the inverse kinematic link are read first. Based on the arrangement order of the leg joint nodes, the leg joint posture transformation relationship, the topological connection relationship, and the four-leg foot end nodes with grounding force attributes, the force transmission path of each leg joint of the four legs is formed. The technical essence of the force transmission path of each leg joint is that the six-degree-of-freedom perturbation vector of the torso enters the leg joint node from the torso centroid reference system and then propagates step by step along the leg joint structure to the foot end nodes of the four legs in a spatial force path. The force transmission path of each leg joint is not simply a geometric connection, but simultaneously includes the arrangement order of the leg joint nodes, the posture transformation relationship of the leg joints, the topological connection relationship, and the grounding force attribute of the foot end nodes. Subsequently, the six-degree-of-freedom perturbation vector of the torso is projected onto the first leg joint node of each force transmission path according to translational and rotational perturbation components, and the initial perturbation component entering the leg joint node is determined based on the posture transformation relationship of that leg joint node. As the initial perturbation component continues to propagate along the force transmission path, at each leg joint node, a perturbation component of the same type is converted according to the posture transformation relationship of the corresponding leg joint and the adjacent topological connection relationship, thus forming the transmitted perturbation component after that leg joint node. To avoid inconsistencies in dimensions during comparison, attenuation feature extraction is performed only within the same perturbation type. For example, the transfer change extraction is performed between translational perturbation components, and between rotational perturbation components, avoiding the direct mixing of translational and rotational perturbation components into a single, meaningless value. Subsequently, the initial perturbation component before each leg joint node or the perturbation component after the previous transfer is extracted with the perturbation component after the transfer of that leg joint node using the same type of amplitude and direction change extraction to form the attenuation coefficient corresponding to that leg joint node. The technical essence of the attenuation coefficient is to express the degree of preservation of the six-degree-of-freedom perturbation vector of the torso as it is transmitted step by step along the force transmission path of each of the four legs, after being jointly changed by the leg joint posture transformation relationship, the topological connection relationship, and the grounding force attributes of the foot end nodes at each leg joint node.The attenuation coefficient can be used to distinguish the transmission influence of different leg joint nodes on the six-degree-of-freedom perturbation vector of the torso, thereby providing a data basis with spatial and force sources for marking the attenuation coefficient as a path feature parameter.

[0039] Preferably, the specific implementation process of the step "marking the attenuation coefficient as the path feature parameter" is as follows: After the attenuation coefficient has been formed along the force transmission path of each leg joint of the four legs, a path feature parameter marking record is first established, and the attenuation coefficient, the six-degree-of-freedom perturbation vector of the torso, the force transmission path of each leg joint of the four legs, the arrangement order of the leg joint nodes, and the foot end nodes of the four legs are written into the path feature parameter marking record. The path feature parameter marking record is used to bind the attenuation coefficient with the corresponding force transmission path, so that the attenuation coefficient is no longer a value that describes the transmission change of a certain leg joint node, but is transformed into a path feature parameter that can characterize the interaction force transmitted from the ends of the arms to the four legs. Subsequently, based on the starting node, the leg joint nodes traversed, and the terminating foot node in the force transmission path of each leg joint, the attenuation coefficients are path-assigned. This path-assignment process ensures that each attenuation coefficient corresponds to a force transmission path from the trunk center of mass reference frame to the foot node, and corresponds to a leg joint node position within that force transmission path. The technical essence of the path feature parameters is to provide a readable, indexable, and associative characteristic representation of the disturbance transmission path in the closed-loop kinematic chain. The path feature parameters include not only the degree of disturbance transmission attenuation reflected by the attenuation coefficient, but also the leg joint node position, the foot node position, and the source of the disturbance type to which the attenuation coefficient belongs. Since the source path of the additional disturbance component experienced by each quadruped foot needs to be determined when constructing the mapping association table, simply retaining the attenuation coefficient cannot directly express which quadruped foot, which force transmission path, and which type of disturbance component the attenuation coefficient belongs to. Therefore, this application marks the attenuation coefficient as the path feature parameter, so that the path feature parameter can simultaneously describe the strength of disturbance transmission and the path location in the subsequent map construction process. Through the binding relationship established by the path feature parameter marking record, the path feature parameter can directly connect the spatial transmission relationship between the six-degree-of-freedom disturbance vector of the torso and the quadruped foot node, so that the subsequent reaction force transmission relationship map can not only express the result of the quadruped foot being disturbed, but also express the process of the disturbance result being transmitted to the quadruped foot node along the closed-loop kinematic chain.

[0040] Preferably, the specific implementation process of the step "using the six-degree-of-freedom perturbation vector of the torso and the path feature parameters as the basis for map construction, constructing a mapping association table recording the additional perturbation components experienced by each quadruped foot end, and then generating the reaction force transmission relationship map based on the mapping association table" is as follows: After the six-degree-of-freedom perturbation vector of the torso and the path feature parameters have been formed, first read each translational perturbation component and each rotational perturbation component in the six-degree-of-freedom perturbation vector of the torso, and read the attenuation coefficient, leg joint node position, force transmission path, and perturbation type source corresponding to each quadruped foot end node in the path feature parameters. Subsequently, path merging processing is performed on the path feature parameters under the same quadruped foot end node to determine the additional perturbation component after the six-degree-of-freedom perturbation vector of the torso reaches the quadruped foot end node through the stress transmission path. The technical essence of the additional disturbance component is that the six-degree-of-freedom disturbance vector of the torso caused by the interaction force at the ends of the two arms at each leg joint is attenuated step by step along the force transmission path of each leg joint and transmitted to the foot end node of the four legs. It describes the incremental disturbance generated by the existing grounding force state of the foot end node of the four legs. The additional disturbance component does not replace the grounding contact force characteristics of each leg, but together with the grounding contact force characteristics of each leg, it characterizes the support and external transmitted disturbances experienced by the foot end node of the four legs in the current control cycle. Subsequently, each foot end node of the four legs, the corresponding additional disturbance component, the source of the six-degree-of-freedom disturbance vector of the torso to which the additional disturbance component belongs, the path characteristic parameters corresponding to the additional disturbance component, and the existing grounding contact force characteristics of each leg of the four legs at the foot end node are written into the mapping association table. The technical essence of the mapping association table is a data structure that stores the six-degree-of-freedom perturbation vector of the torso, the path feature parameters, the additional perturbation components, and the ground contact force characteristics of each leg according to the four-legged foot-end nodes. In the mapping association table, each row corresponds to a four-legged foot-end node, and each column records the force transmission path, path feature parameters, additional perturbation components, and ground contact force characteristics of each leg associated with that four-legged foot-end node. This ensures that the additional perturbation components experienced by each four-legged foot-end node can be traced back to both the six-degree-of-freedom perturbation vector of the torso and the path feature parameters. When generating the reaction force transmission relationship map based on the mapping association table, the end nodes of both arms, the torso centroid reference system, the leg joint nodes, and the four-legged foot-end nodes are used as map nodes. The force transmission paths corresponding to the path feature parameters are used as map connection relationships. The additional perturbation components and the ground contact force characteristics of each leg are configured into the corresponding four-legged foot-end nodes to form the reaction force transmission relationship map.The technical essence of the reaction force transmission relationship map is to further add a data-driven expression of the strength of disturbance transmission and the results of disturbance at the ends of the four legs on the basis of the topological connection relationship of the closed-loop kinematic chain. This enables the differential effects of the interaction force on the ends of the two arms on different nodes of the four legs when the initial stiffness parameters are compensated based on the reaction force transmission relationship map.

[0041] Optionally, the step of parsing the single-leg swing marker or full-support marker carried in the quadrupedal support phase parameters to identify the quadrupedal support state of the centaur robot, and configuring initial stiffness parameters for the ends of the twin arms based on the single-leg swing marker or full-support marker, includes: The single-leg swing marker carried in the quadrupedal support phase parameter is analyzed to detect whether the quadrupedal support state of the centaur robot corresponds to the single-leg swing state. When the single-leg swing marker is detected, the centaur robot is determined to be in a walking state or a posture adjustment state based on the single-leg swing marker. Extract the first impedance reference profile stored in the first data pool, the first impedance reference profile containing the set compliance follow stiffness value; Using the set compliant following stiffness value as the stiffness source corresponding to the single-leg swing mark, the set compliant following stiffness value is assigned to the ends of both arms to generate the initial stiffness parameter corresponding to the single-leg swing mark, thereby enabling the ends of both arms to have compliant characteristics that absorb external vibrations.

[0042] Preferably, the specific implementation process of the step "parse the single-leg swing marker carried in the quadrupedal support phase parameter to detect whether the quadrupedal support state of the centaur robot corresponds to the single-leg swing state" is as follows: After the quadrupedal support phase parameter has been constructed by the support period marker and swing period marker corresponding to each quadrupedal foot, the current control cycle marker, foot support state field, support period marker, swing period marker, and quadrupedal foot identifier in the quadrupedal support phase parameter are read first, and the current control cycle marker, the foot support state field, the support period marker, the swing period marker, and the quadrupedal foot identifier are written into the support state parsing record. The support state parsing record is used to associate the foot support state field and the quadrupedal foot identifier of each quadrupedal foot in the current control cycle, so that the single-leg swing marker is not an independent identifier detached from the foot support state field, but can be traced back to the foot support state field in the quadrupedal support phase parameter. Subsequently, based on the foot end identifiers of the four legs, the support phase markers and swing phase markers in the support state analysis record are checked for foot end count to determine the number of four legs in the support phase and the number of four legs in the swing phase. This foot end count check does not change the support phase markers and swing phase markers in the four-leg support phase parameters; instead, it converts the support phase markers and swing phase markers scattered under different four-leg foot end identifiers into support state determination data that can express the overall machine support configuration. Afterward, the support state determination data is matched with the single-leg swing state determination caliber. The single-leg swing state determination caliber is written into the support state determination configuration record during the basic gait parameter configuration stage. The support state determination configuration record includes the total number of four legs' foot ends, the number of swing phase foot ends corresponding to the single-leg swing state, the number of support phase foot ends corresponding to the single-leg swing state, and the reading method of the current control cycle marker. The basic gait parameter configuration stage is used to pre-write the total number of quadrupedal feet, the support phase marker generation method, and the swing phase marker generation method of the Centaur robot into the support state determination configuration record. This ensures that the support state determination configuration record provides a consistent source of determination for the support state determination data within the current control cycle. The support state determination configuration record is not an additional, isolated configuration but rather serves to explain how to identify the single-leg swing marker from the quadrupedal support phase parameters. This allows the same Centaur robot to analyze the quadrupedal support state according to the same single-leg swing state determination caliber across different gait cycles. When the support state determination data indicates that only one quadrupedal foot is in the swing phase while the other four quadrupedal feet are in the support phase, the foot combination state corresponding to the current control cycle marker is determined as the single-leg swing marker. The technical essence of the single-leg swing marker is a data-driven representation of the support configuration in the quadrupedal support phase parameters where one quadrupedal foot is detached from ground support while the other four quadrupedal feet remain grounded.The single-leg swing marker can be used to elevate the combination of the support phase marker and the swing phase marker on the four legs to the basis for identifying the overall four-leg support state of the robot, providing a direct data source for subsequent determination of whether the Centaur robot is in a walking or posture adjustment state based on the single-leg swing marker.

[0043] Preferably, in the specific technical implementation of the step "when the single-leg swing mark is detected, determine whether the centaur robot is in a walking state or a posture adjustment state based on the single-leg swing mark", after the single-leg swing mark has been identified by the support state parsing record, the current control cycle mark, the swing phase quadruped foot end mark, the support phase quadruped foot end mark, the swing phase mark duration, and the posture angle change trend of the robot's torso corresponding to the single-leg swing mark are read, and the current control cycle mark, the swing phase quadruped foot end mark, the support phase quadruped foot end mark, the swing phase mark duration, and the posture angle change trend are written into the single-leg swing state determination record. The single-leg swing state determination record is used to further distinguish the single-leg swing mark into a walking state or a posture adjustment state, rather than classifying all single-leg swing states into the same operating condition. Subsequently, based on the switching order of the quadruped foot markers during the swing phase in a continuous control cycle, the single-leg swing state determination record undergoes foot rotation recognition processing. This foot rotation recognition processing sequentially compares the quadruped foot markers during the swing phase, the quadruped foot markers during the support phase, and the duration of the swing phase markers within the continuous control cycle to form the basis for determining the single-leg swing state. This single-leg swing state determination basis is further correlated with the robot's torso posture angle change trend, ensuring that the determination of the walking state or the posture adjustment state originates not only from the changes in the quadruped foot markers during the swing phase but also from the posture changes of the robot's torso within the current control cycle. When the quadruped foot markers during the swing phase rotate along the gait sequence of the quadruped mechanism within a continuous control cycle, and the posture angle change trend falls within the torso posture change range corresponding to the walking process, the centaur robot is determined to be in a walking state. The torso posture change range is written into the support state determination configuration record along with the single-leg swing state determination criteria during the basic gait parameter configuration stage, and is used in the single-leg swing state determination record to verify the posture angle change trend. The technical essence of the traveling state is that the quadruped mechanism propels the robot's torso to displacement by sequentially entering the swing phase with different foot ends. This traveling state causes robot torso movement relative to the ends of the arms. Therefore, the traveling state needs to participate in the subsequent initial stiffness parameter configuration so that the ends of the arms do not use the higher stiffness configuration of the standing operation state when making contact. On the other hand, when the foot end markers during the swing phase remain at the same foot end within a continuous control cycle, or when the foot end markers during the swing phase undergo a small range of change around the foot end combination formed by the foot end markers during the swing phase and the support phase, and the trend of the attitude angle change reflects the adjustment needs of the robot's torso attitude, the Centaur robot is determined to be in an attitude adjustment state.The technical essence of the posture adjustment state is that the quadruped mechanism triggers a single quadruped foot to enter a swinging phase in order to correct the robot's torso posture angle or the rate of change of posture angle. This posture adjustment state also affects the spatial positional stability of the end effectors through robot body disturbances. Therefore, both the traversal state and the posture adjustment state originate from the single-leg swing marker, but their corresponding operational meanings are different. By unifying the traversal state and the posture adjustment state into the initial stiffness parameter configuration branch corresponding to the single-leg swing marker, the end effectors can use a set compliant following stiffness value as their stiffness source when in a single-leg swinging state under quadrupedal support, thereby reducing the direct transmission of robot body disturbances to the end effectors' contact operations.

[0044] Preferably, the specific implementation process of the step "extracting the first impedance reference configuration file stored in the first data pool, the first impedance reference configuration file containing the set compliant following stiffness value" is as follows: After the centaur robot is determined to be in the walking state or the posture adjustment state, a first impedance reference reading record is first formed based on the current control cycle marker, the single-leg swing marker, the walking state, or the posture adjustment state, and the first impedance reference reading record is used to access the first data pool. The first data pool is established when the centaur robot completes assembly calibration and basic control parameter configuration. The first data pool stores impedance reference configuration files under different support conditions according to the four-leg support state category; wherein, the impedance reference configuration file corresponding to the single-leg swing marker is written as the first impedance reference configuration file, and the first impedance reference configuration file establishes a correspondence with the single-leg swing marker, the walking state, and the posture adjustment state. The technical essence of the first data pool is a data storage area used to store the mapping relationship between the four-leg support state and the impedance configuration of the two-arm end caps. The first data pool is not a simple parameter table, but rather associates the single-leg swing marker identified in the current control cycle with the corresponding first impedance reference configuration file through the first impedance reference reading record. The first impedance reference configuration file includes a compliant following stiffness value field, a compliant following damping value field, a dual-arm end direction field, and an applicable support state field; wherein, the compliant following stiffness value field records the set compliant following stiffness value, the compliant following damping value field records the damping reference that matches the set compliant following stiffness value, the applicable support state field records the traveling state or attitude adjustment state corresponding to the single-leg swing mark, and the dual-arm end direction field is used to distinguish the stiffness configuration caliber of the dual-arm end in different spatial directions. The set compliant following stiffness value is configured during the basic control parameter configuration based on the allowable following displacement range of the dual-arm end-effectors, the expected contact force range during contact operation, and the robot body disturbance range when the quadruped mechanism is in a single-leg swinging state. The allowable following displacement range of the dual-arm end-effectors limits the finite displacement that the dual-arm end-effectors can undergo under robot body disturbance. The expected contact force range during contact operation limits the force response range of the dual-arm end-effectors during contact operation. The robot body disturbance range limits the range of position and posture changes of the robot torso relative to the dual-arm end-effectors in the traveling or posture adjustment states. This configuration process does not directly superimpose the robot body disturbance in the traveling or posture adjustment states onto the dual-arm end-effectors, but instead reserves compliant following space for the dual-arm end-effectors through a lower stiffness caliber.Subsequently, the set compliant following stiffness value is extracted from the first impedance reference configuration file and written into the first impedance reference read record, so that the set compliant following stiffness value can continue to participate in the subsequent processing of configuring the initial stiffness parameters of the dual arm end caps. The technical essence of the set compliant following stiffness value is to provide a reference value with a lower resistance stiffness for the dual arm end caps in the single-leg swing state corresponding to the single-leg swing mark, so that the dual arm end caps can generate a limited following displacement along the direction corresponding to the dual arm end cap direction field when subjected to robot body disturbance, instead of transmitting all robot body disturbance to the dual arm end cap contact point with higher stiffness.

[0045] Preferably, the specific implementation process of the step "using the set compliant following stiffness value as the stiffness source corresponding to the single-leg swing mark, assigning the set compliant following stiffness value to the ends of both arms to generate the initial stiffness parameter corresponding to the single-leg swing mark, thereby enabling the ends of both arms to have compliant characteristics that absorb external vibrations" is as follows: After the set compliant following stiffness value has been extracted from the first impedance reference configuration file, the set compliant following stiffness value, the single-leg swing mark, the walking state or the posture adjustment state, the direction field of the ends of both arms, and the current control cycle mark are read from the first impedance reference reading record, and the set compliant following stiffness value, the single-leg swing mark, the walking state or the posture adjustment state, the direction field of the ends of both arms, and the current control cycle mark are written into the initial stiffness parameter generation record. The initial stiffness parameter generation record is used to uniformly record the source of the set compliant following stiffness value, the source of the support state of the single-leg swing mark, and the configuration direction of the two arm ends, so that the initial stiffness parameter can be traced back to the four-leg support state within the current control cycle. Subsequently, based on the direction field of the two arm ends, the set compliant following stiffness value is configured to the corresponding direction of the two arm ends in the torso center of mass reference system; when the direction field of the two arm ends includes the forward direction, the lateral direction, and the vertical direction, the set compliant following stiffness value enters the stiffness configuration position of the corresponding direction, so as to form the directional stiffness configuration result of the two arm ends. The directional stiffness configuration result of the two arm ends does not change the set compliant following stiffness value itself, but rather maps the set compliant following stiffness value to the directional relationship of the two arm ends under a unified spatial reference, so that the two arm ends can retain compliant following space in the direction more sensitive to robot body disturbances when in the traveling state or posture adjustment state. Subsequently, the directional stiffness configuration result of the dual arm end caps is written into the initial stiffness parameter generation record to generate the initial stiffness parameter corresponding to the single-leg swing mark. The technical essence of the initial stiffness parameter is that, given the single-leg swing mark indicates a quadrupedal support state with a single-leg swing, it sets the initial impedance value for the dual arm end caps. The initial stiffness parameter will subsequently undergo compensation processing based on the reaction force transmission relationship graph; therefore, the initial stiffness parameter is not the final parameter used to drive the dual arm joints, but rather the stiffness input basis before the formation of the target force-position impedance adjustment parameter. By using the set compliant following stiffness value as the stiffness source corresponding to the single-leg swing mark, when the dual arm end caps are subjected to robot body disturbances caused by the robot's torso during movement or posture adjustment, they can form limited displacement following through a lower stiffness configuration, thereby enabling the dual arm end caps to possess compliant characteristics that absorb robot body disturbances.The technical essence of the aforementioned compliance characteristic is to reduce the resistance of the end arms to instantaneous positional deviations, thereby weakening the direct transmission of robot body interference motion to the contact points of the end arms. This allows for subsequent compensation processing based on the reaction force transmission relationship diagram to further combine the overall machine force distribution characteristics with the existing initial stiffness parameters to form target force-potential impedance adjustment parameters.

[0046] Optionally, the step of parsing the single-leg swing marker or full-support marker carried in the quadrupedal support phase parameters to identify the quadrupedal support state of the centaur robot, and configuring initial stiffness parameters for the ends of the twin arms based on the single-leg swing marker or full-support marker, further includes: The full support marker carried in the quadrupedal support phase parameter is analyzed to detect whether the quadrupedal support state of the centaur robot corresponds to the full support state. When the full support marker is detected, it is determined that the Centaur robot is in a stationary working state based on the full support marker; Extract the second impedance reference profile stored in the second data pool. The second impedance reference profile contains positioning maintenance stiffness values ​​configured for positioning operations. The positioning maintenance stiffness values ​​have stronger disturbance rejection properties compared to the compliant following stiffness values. Using the positioning and maintaining stiffness value as the stiffness source corresponding to the full support mark, the positioning and maintaining stiffness value is assigned to the end of the dual arms to generate the initial stiffness parameter corresponding to the full support mark, thereby controlling the positioning accuracy of the end of the dual arms when performing operations.

[0047] Preferably, the specific implementation process of the step "parse the full support marker carried in the quadrupedal support phase parameter to detect whether the quadrupedal support state of the Centaur robot corresponds to the full support state" is as follows: After the quadrupedal support phase parameter has been constructed by the support period marker and swing period marker corresponding to each quadrupedal foot, the current control cycle marker, foot support state field, support period marker, swing period marker, and quadrupedal foot identifier in the quadrupedal support phase parameter are read first, and the current control cycle marker, the foot support state field, the support period marker, the swing period marker, and the quadrupedal foot identifier are written into the full support state parsing record. The full support state parsing record is used to associate the foot support state field corresponding to each quadrupedal foot in the current control cycle with the quadrupedal foot identifier, so that the full support marker can be parsed from the support period marker and swing period marker in the quadrupedal support phase parameter, rather than as an independent identifier detached from the quadrupedal support phase parameter. Subsequently, the support phase markers and swing phase markers in the full support state analysis record are checked for support integrity based on the quadruped foot end markers to form full support state determination data. This full support state determination data includes the number of quadruped feet in the support phase, the number of quadruped feet in the swing phase, and the foot end support state field corresponding to each quadruped foot end marker. The support integrity check does not change the support phase markers and swing phase markers in the quadruped support phase parameters; instead, it converts the support phase markers and swing phase markers scattered under each quadruped foot end marker into full support state determination data that can express the full support configuration. Afterward, the full support state determination data is matched with the full support state determination caliber. The full support state determination caliber is written into the support state determination configuration record during the basic gait parameter configuration stage. This support state determination configuration record includes the total number of quadruped feet, the number of feet in the support phase corresponding to the full support state, the number of feet in the swing phase corresponding to the full support state, and the reading method of the current control cycle marker. The basic gait parameter configuration stage is used to write the total number of quadrupedal foot ends, the method for generating support phase markers, and the method for generating swing phase markers into the support state determination configuration record, so that the support state determination configuration record can provide a common determination source for the full support state determination data. The support state determination configuration record is used to provide the full support state determination caliber in different control cycles, so that the full support state determination data can parse the full support markers according to the full support state determination caliber. When the full support state determination data indicates that all quadrupedal foot ends are in the support phase and there are no quadrupedal foot ends in the swing phase, the foot end combination state corresponding to the current control cycle marker is determined as the full support marker.The technical essence of the full support marker is the data representation of the full support configuration in which all four legs maintain ground support in the four-leg support phase parameter. Through the full support marker, the local support states of multiple four-legged ends can be merged into the identification basis of the overall four-leg support state, providing a data source for subsequent determination of the Centaur robot's standing operation state based on the full support marker.

[0048] Preferably, in the specific technical implementation of the step "when the full support mark is detected, determine that the Centaur robot is in a standing operation state based on the full support mark", after the full support mark has been identified by the full support state parsing record, the current control cycle mark corresponding to the full support mark, the foot end markers of each quadruped, the support period markers corresponding to each foot end marker, the duration of the full support mark, and the attitude angle change trend of the robot's torso are read and written into the standing operation state determination record. The standing operation state determination record is used to further convert the full support mark into a standing operation state, rather than just stopping at the state identification level of whether the feet are grounded. Subsequently, based on the duration of the full support markers, the standing operation state determination record undergoes a support continuity check. This check determines whether the full support markers persist throughout the continuous control cycle and writes the support period markers corresponding to each quadruped foot marker within the continuous control cycle into the standing operation state determination record, enabling the record to express the time relationship of all quadruped feet being continuously grounded. Next, the trend of attitude angle changes is matched with the range of torso attitude changes corresponding to the standing operation state. The range of torso attitude changes is written into the support state determination configuration record along with the full support state determination criteria during the basic gait parameter configuration stage and is used in the standing operation state determination record to verify whether the robot's torso is in a standing attitude change state. When the full support markers persist throughout the continuous control cycle and the trend of attitude angle changes falls within the range of torso attitude changes, the Centaur robot is determined to be in a standing operation state. The technical essence of the standing operation state is that the quadruped mechanism provides ground support through the joint of all four foot ends, keeping the robot's torso in a standing posture change state during the current control cycle. This standing operation state differs from the aforementioned traveling or posture adjustment states. In this state, the robot's torso experiences less disturbance, making it more suitable for the end arms to use higher stiffness calibers to maintain the contact operation position. Therefore, the full support marker not only indicates the number of foot ends grounded but also triggers the initial stiffness parameter configuration branch corresponding to the standing operation state. This allows the end arms to use a positioning maintenance stiffness value as the stiffness source in the full support state, providing an impedance configuration basis for subsequent positioning operations. This impedance configuration basis continues to participate in the subsequent processing of configuring the initial stiffness parameters of the end arms, creating a continuous technical interaction between the standing operation state, the positioning maintenance stiffness value, and the initial stiffness parameters.

[0049] Preferably, the specific implementation process of the step "extracting the second impedance reference configuration file stored in the second data pool, the second impedance reference configuration file containing positioning maintenance stiffness values ​​configured for positioning operations, the positioning maintenance stiffness values ​​having stronger anti-disturbance properties compared to the compliant following stiffness values" is as follows: After the Centaur robot is determined to be in the standing operation state, a second impedance reference reading record is first formed based on the current control cycle marker, the full support marker, and the standing operation state, and the second impedance reference reading record is used to access the second data pool. The second data pool is established when the Centaur robot completes assembly calibration and basic control parameter configuration, and the second data pool stores impedance reference configuration files according to the positioning operation requirements in the full support state; wherein, the impedance reference configuration file corresponding to the full support marker and the standing operation state is written as the second impedance reference configuration file. The second data pool is essentially a data storage area for storing the mapping relationship between the full-support state and the impedance configuration of the dual-arm end-positioning operation. The second data pool differs from the first data pool, which stores the compliant following stiffness value in the single-leg swing state, while the second data pool stores the positioning maintenance stiffness value in the full-support state. Therefore, the reading conditions of the second data pool originate from the full-support marker and the standing operation state. The second impedance reference configuration file includes a positioning maintenance stiffness value field, a positioning maintenance damping value field, a dual-arm end-positioning direction field, and an applicable support state field. The positioning maintenance stiffness value field records the positioning maintenance stiffness value; the positioning maintenance damping value field records the positioning maintenance damping value corresponding to the positioning maintenance stiffness value; the dual-arm end-positioning direction field distinguishes the stiffness configuration aperture of the dual-arm end-positioning in different directions within the torso center-of-gravity reference system; and the applicable support state field records the standing operation state corresponding to the full-support marker. The positioning and maintaining stiffness value is configured in the basic control parameter configuration based on the allowable position deviation range of the dual-arm end-effector positioning operation, the expected contact force range when the dual-arm end-effectors are in contact operation, and the robot body disturbance range in the fully supported state. The allowable position deviation range of the dual-arm end-effector positioning operation is used to limit the displacement deviation that the dual-arm end-effectors can cause in the stationary operation state. The expected contact force range when the dual-arm end-effectors are in contact operation is used to limit the force response range of the dual-arm end-effectors in the positioning operation. The robot body disturbance range in the fully supported state is used to limit the range of position and attitude changes of the robot body relative to the dual-arm end-effectors when all four feet are in contact with the ground.Since the positioning maintenance stiffness value is geared towards positioning operations in the stationary working state, it exhibits stronger disturbance resistance compared to the compliant following stiffness value. The technical essence of this disturbance resistance is that the end effectors of the dual arms, when subjected to robot body disturbances or contact work disturbances in a fully supported state, exhibit a higher degree of resistance to instantaneous position deviations, thereby minimizing the range of spatial position changes during positioning operations. Subsequently, the positioning maintenance stiffness value is extracted from the second impedance reference configuration file and written into the second impedance reference read record, allowing it to continue participating in subsequent processing for configuring the initial stiffness parameters of the dual arms. The second impedance reference read record also records the source of the second impedance reference configuration file, enabling the positioning maintenance stiffness value to be traced back to the second data pool, the fully supported marker, and the stationary working state.

[0050] Preferably, the specific implementation process of the step "using the positioning and maintaining stiffness value as the stiffness source corresponding to the full support mark, assigning the positioning and maintaining stiffness value to the dual-arm end effector to generate the initial stiffness parameter corresponding to the full support mark, thereby controlling the positioning accuracy of the dual-arm end effector when performing operation" is as follows: After the positioning and maintaining stiffness value has been extracted from the second impedance reference configuration file, the positioning and maintaining stiffness value, the full support mark, the stationing operation state, the dual-arm end effector direction field, and the current control cycle mark are first read from the second impedance reference reading record, and the positioning and maintaining stiffness value, the full support mark, the stationing operation state, the dual-arm end effector direction field, and the current control cycle mark are written into the full support initial stiffness parameter generation record. The full support initial stiffness parameter generation record is used to uniformly record the source of the positioning and maintaining stiffness value, the source of the support state of the full support mark, and the configuration direction of the dual-arm end effector, so that the initial stiffness parameter corresponding to the full support mark can be traced back to the stationing operation state within the current control cycle. Subsequently, based on the direction field of the dual-arm end caps, the positioning and maintaining stiffness values ​​are configured to the corresponding directions of the dual-arm end caps in the torso center of mass reference system. When the dual-arm end cap direction field includes the forward direction, the lateral direction, and the vertical direction, the positioning and maintaining stiffness values ​​are respectively entered into the stiffness configuration positions in the corresponding directions to form a directional positioning stiffness configuration result for the dual-arm end caps. The directional positioning stiffness configuration result for the dual-arm end caps does not change the positioning and maintaining stiffness values ​​themselves, but rather maps the positioning and maintaining stiffness values ​​to the directional relationship of the dual-arm end caps under a unified spatial reference, enabling the dual-arm end caps to maintain high resistance stiffness along the directions most sensitive to positioning and operation in the stationary operation state. Afterward, the directional positioning stiffness configuration result for the dual-arm end caps is written into the initial stiffness parameter generation record of the full support to generate the initial stiffness parameters corresponding to the full support marker. The technical essence of the initial stiffness parameter corresponding to the full support mark is that it is the initial impedance value configured for the ends of the dual arms when the full support mark indicates that the four-legged support state is in a fully supported state. The initial stiffness parameter corresponding to the full support mark will subsequently undergo compensation processing based on the reaction force transmission relationship diagram. Therefore, the initial stiffness parameter corresponding to the full support mark is not the final parameter used to drive the dual arm joints, but rather the stiffness input basis before the formation of the target force-position impedance adjustment parameter. By using the positioning maintenance stiffness value as the stiffness source corresponding to the full support mark, when the ends of the dual arms are subjected to contact operation disturbances in the stationary operation state, the displacement response corresponding to the instantaneous position deviation can be reduced through a higher stiffness configuration, thereby controlling the positioning accuracy of the ends of the dual arms when performing operation.The technical essence of the positioning accuracy is the degree of spatial deviation of the end arms relative to the target working position during the positioning operation. By using the initial stiffness parameter corresponding to the full support mark, the target force-position impedance adjustment parameter can be formed by combining the force distribution characteristics of the whole machine on the basis of the existing positioning maintenance stiffness value when subsequent compensation processing is carried out based on the reaction force transmission relationship diagram.

[0051] Optionally, the step of compensating the initial stiffness parameter based on the reaction force transmission relationship diagram to generate target potential impedance adjustment parameters for the ends of the double arms includes: Extract the overall machine stiffness coupling matrix under the current attitude from the reaction force transmission relationship graph; Locate the stiffness element corresponding to the supporting foot in the overall machine stiffness coupling matrix; The stiffness element of the support foot is compared with a set standing threshold to determine whether the stiffness element of the support foot is lower than the set standing threshold. If the stiffness element of the support foot is lower than the set standing threshold, then the deviation characteristic value of the set standing threshold relative to the stiffness element of the support foot is calculated. A stiffness attenuation compensation amount is generated based on the deviation characteristic value; The initial stiffness parameter is reduced in order by using the stiffness attenuation compensation amount to generate the corrected impedance parameter, and the corrected impedance parameter is determined as the target force potential impedance adjustment parameter for the ends of the two arms.

[0052] Preferably, the specific implementation process of step "extracting the overall stiffness coupling matrix of the current posture from the reaction force transmission relationship graph" is as follows: After the reaction force transmission relationship graph has been generated based on the mapping association table, the graph nodes, graph connection relationships, path feature parameters, additional disturbance components, and ground contact force characteristics of each leg in the reaction force transmission relationship graph are read first, and the robot torso posture relationship expressed by the torso centroid reference system in the current control cycle is read simultaneously to establish a current posture stiffness extraction record. The current posture stiffness extraction record is used to incorporate the path feature parameters, additional disturbance components, ground contact force characteristics of each leg, and robot torso posture relationship in the reaction force transmission relationship graph into the same stiffness evaluation caliber, so that the subsequently formed overall stiffness coupling matrix can correspond to the current control cycle, rather than to the static assembly state. Subsequently, based on the dual-arm end-effector nodes, torso center-of-gravity reference system, leg joint nodes, and quadruped end-effector nodes in the graph nodes, the row and column organization of the overall machine stiffness coupling matrix is ​​determined. The rows of the overall machine stiffness coupling matrix correspond to the impedance adjustment directions of the dual-arm end-effectors in the forward, lateral, and vertical directions under the torso center-of-gravity reference system. The columns correspond to the quadruped end-effector support positions formed when each quadruped end-effector provides support to the robot torso. The matrix elements at the intersection of rows and columns express the support constraint capability of the corresponding quadruped end-effector node for adjusting the force position of the dual-arm end-effectors in the corresponding impedance adjustment direction. Next, the disturbance transmission attenuation degree expressed by the path feature parameters, the quadruped end-effector disturbance increment expressed by the additional disturbance components, and the grounding force basis expressed by the grounding contact force characteristics of each leg are processed to form the current attitude stiffness contribution data. The current attitude stiffness contribution data does not directly mix forces, displacements, and attitude angles of different dimensions for calculation. Instead, it first assigns data according to the four-legged foot-end nodes, impedance adjustment direction, and current control cycle. Then, it converts the disturbance transmission attenuation degree, the disturbance increment at the four-legged foot-ends, and the grounding force foundation under the same assignment into contribution relationships under the same stiffness evaluation caliber. Finally, the current attitude stiffness contribution data is written into the matrix element positions corresponding to the four-legged foot-end nodes and impedance adjustment direction in the overall stiffness coupling matrix to form the overall stiffness coupling matrix under the current attitude. The technical essence of the overall stiffness coupling matrix is ​​to organize the force coupling relationships scattered on different force transmission paths in the reaction force transmission relationship diagram into a data structure that can participate in the initial stiffness parameter compensation processing. This allows for the subsequent determination of the support constraint capability of different support legs on the impedance configuration of the dual-arm ends, thus ensuring that the generation of the target force-position impedance adjustment parameters no longer depends solely on the pre-configured stiffness source.

[0053] Preferably, in the specific technical implementation of the step "locating the stiffness element of the corresponding support foot in the overall stiffness coupling matrix", after the overall stiffness coupling matrix has been formed, the foot support state field, support period mark, swing period mark, and four-leg foot identifier in the four-leg support phase parameters are read first, and the foot support state field, the support period mark, the swing period mark, and the four-leg foot identifier are written into the support foot stiffness element positioning record. The support foot stiffness element positioning record is used to correspond the matrix elements in the overall stiffness coupling matrix with the four-leg feet in the support period in the current control cycle, avoiding the matrix elements corresponding to the four-leg feet in the swing period as support foot stiffness elements. Subsequently, the four-leg feet in the support period are filtered according to the support period mark, and the four-leg foot identifier corresponding to the four-leg feet in the support period is determined as the support foot identifier; the support foot identifier is further processed to correspond with the four-leg foot nodes in the reaction force transmission relationship diagram to determine the column position of the support foot identifier in the overall stiffness coupling matrix. Simultaneously, the direction position requiring impedance adjustment is determined based on the direction field at the ends of the two arms. This direction field originates from the aforementioned first or second impedance reference configuration file and is used to express the stiffness configuration direction of the ends of the two arms within the torso center of mass reference system. By cross-locating the column position corresponding to the support foot identifier and the row position corresponding to the direction field at the ends of the two arms, the stiffness element of the corresponding support foot can be located in the overall stiffness coupling matrix. The technical essence of the stiffness element of the support foot is the support constraint capability of a certain foot end in the current posture and current support state on a certain impedance adjustment direction of the ends of the two arms. This support constraint capability originates from the ground contact force characteristics of each leg and is also affected by the path characteristic parameters and the additional disturbance components. Therefore, the stiffness element of the support foot is not a simple foot contact pressure value, nor is it a single stiffness value at the ends of the two arms, but rather a readable expression of the reaction force transmission relationship spectrum in the overall stiffness coupling matrix. By locating the stiffness element of the corresponding support foot in the overall stiffness coupling matrix, it is possible to determine whether the support constraint capacity of the four feet that are in the support period and bear the support function is insufficient, thereby providing a clear data source for setting the standing threshold comparison, calculating the deviation characteristic value, and generating the stiffness attenuation compensation amount.

[0054] Preferably, the specific implementation process of the step "comparing the stiffness element of the support foot with the set stationing threshold to determine whether the stiffness element of the support foot is lower than the set stationing threshold" is as follows: After the stiffness element of the support foot has been located by the support foot stiffness element positioning record, the stationing threshold configuration record is read first, and the set stationing threshold is read from the stationing threshold configuration record. The stationing threshold configuration record is formed in the basic control parameter configuration stage. The stationing threshold configuration record includes the dual-arm end direction field, the minimum support stiffness caliber corresponding to the stationing operation state, the support period mark reading method, and the current control cycle mark matching method. The basic control parameter configuration stage is used to write the correspondence between the dual-arm end impedance adjustment direction and the minimum support constraint capability of the four legs in the stationing operation state into the stationing threshold configuration record, so that the set stationing threshold can be read according to the dual-arm end direction field and the current control cycle mark. The set stationing threshold is not an arbitrary empirical value, but is used to express the minimum support constraint capability that the four legs should provide to the dual-arm end impedance adjustment direction under the support conditions of the stationing operation state or close to the stationing operation state. Since the initial stiffness parameter in this application may originate from the set compliant following stiffness value or the positioning maintenance stiffness value, if the current support foot's support constraint capability for the ends of the two arms is insufficient, and a higher initial stiffness parameter is still directly applied to the ends of the two arms, the reaction force generated when the ends of the two arms are subjected to interactive forces can easily be further transmitted to the four legs with insufficient support constraint capability. Therefore, it is necessary to compare the stiffness element of the support foot with the set standing threshold to determine whether the support foot has the support constraint capability to bear the current initial stiffness parameter. Specifically, during the comparison, firstly, based on the direction field of the ends of the two arms, select a set standing threshold from the standing threshold configuration record that belongs to the same impedance adjustment direction as the stiffness element of the support foot, and then check whether the stiffness element of the support foot and the set standing threshold are in the same stiffness evaluation caliber to avoid direct comparison of values ​​in different directions or under different support states. After verifying the same stiffness evaluation criteria, the stiffness element of the support foot is compared with the set standing threshold. When the stiffness element of the support foot is lower than the set standing threshold, it indicates that the support foot's ability to support and constrain the impedance adjustment direction of the dual arms in the current posture is lower than the minimum support and constraining ability corresponding to the standing operation state. The comparison result is subsequently used in the deviation characteristic value calculation, ensuring that the initial stiffness parameter does not deviate from the support constraint capability of the support foot and directly enters the target force potential impedance adjustment parameter generation process.

[0055] Preferably, the specific implementation process of step "if the stiffness element of the support foot is lower than the set standing threshold, then calculate the deviation characteristic value of the set standing threshold relative to the stiffness element of the support foot" is as follows: When the stiffness element of the support foot is lower than the set standing threshold, a deviation characteristic value calculation record is first established, and the stiffness element of the support foot, the set standing threshold, the support foot identifier, the direction field of the double arm end, the current control cycle marker, and the comparison result are written into the deviation characteristic value calculation record. The deviation characteristic value calculation record is used to bind the source of insufficient stiffness of the support foot, the direction of insufficient stiffness of the support foot, and the current control cycle, so that the deviation characteristic value formed subsequently can be traced back to which support foot, which impedance adjustment direction, and which set standing threshold comparison. Subsequently, in the deviation characteristic value calculation record, the difference extraction processing under the same stiffness evaluation caliber is performed on the set standing threshold and the stiffness element of the support foot to form a stiffness deficiency difference quantity; the stiffness deficiency difference quantity is used to express the deficiency magnitude of the stiffness element of the support foot relative to the set standing threshold. To enable the stiffness deficiency difference to further participate in the generation of subsequent stiffness attenuation compensation, a standardized conversion is needed in conjunction with the set stationing threshold to form deviation characteristic values. The technical function of this standardized conversion is to convert the stiffness deficiency differences under different directions, different support feet, and different control cycles into comparable degrees of deviation, allowing the same compensation standard to be used when generating subsequent stiffness attenuation compensation. The technical essence of the deviation characteristic value is to describe the degree to which the stiffness element of the support foot is lower than the set stationing threshold, rather than simply recording the numerical difference between the set stationing threshold and the stiffness element of the support foot. Since the deviation characteristic value is simultaneously bound to the support foot identifier, the dual-arm end direction field, and the current control cycle marker, it can correspond to a specific initial stiffness parameter correction direction in subsequent processing. If multiple support feet have stiffness elements lower than the set stationing threshold within the same current control cycle, multiple deviation characteristic values ​​are generated, and the correspondence between each deviation characteristic value and the support foot identifier is retained in the deviation characteristic value calculation record. This avoids losing the direction and foot end source after simply averaging multiple support foot stiffness deficiency cases. Based on the deviation eigenvalue, the subsequent stiffness attenuation compensation can be generated according to the degree of insufficient stiffness of the support foot, so that the order reduction correction of the initial stiffness parameter corresponds to the support constraint capacity in the reaction force transmission relationship diagram.

[0056] Preferably, in the specific technical implementation of the step "generating stiffness attenuation compensation amount based on the deviation feature value", after the deviation feature value has been written into the deviation feature value calculation record, the deviation feature value, the support foot identifier, the direction field of the two arm ends, the current control cycle marker, and the path feature parameter corresponding to the support foot identifier are read, and the deviation feature value, the support foot identifier, the direction field of the two arm ends, the current control cycle marker, and the path feature parameter are written into the stiffness attenuation compensation amount generation record. The stiffness attenuation compensation amount generation record is used to correspond the deviation feature value with the force transmission path from the two arm ends to the support foot due to the interaction force, so that the stiffness attenuation compensation amount not only reflects the degree of insufficient stiffness of the support foot, but also reflects the path correlation of the degree of insufficient stiffness of the support foot when the two arm ends are subjected to the interaction force transmitted to the four feet. Subsequently, an initial attenuation ratio is determined based on the deviation characteristic value, and the initial attenuation ratio is path-correlatedly corrected based on the path characteristic parameters to form a path-correlated attenuation ratio. The essence of this path-correlated correction is that when the stiffness element of a support foot is lower than the set standing threshold and the support foot is simultaneously on the main transmission path of the interaction force at the ends of the two arms, the attenuation degree in the corresponding direction of the support foot is increased; conversely, the attenuation degree in the corresponding direction of the support foot is decreased. After completing the path-correlated correction, the path-correlated attenuation ratio is processed in correspondence with the stiffness component of the field corresponding to the ends of the two arms in the initial stiffness parameters to form a stiffness attenuation compensation amount. The essence of this stiffness attenuation compensation amount is the stiffness reduction range determined jointly by the degree of stiffness deficiency of the support foot and the force transmission path from the ends of the two arms to the support foot. The stiffness attenuation compensation amount does not unconditionally weaken the initial stiffness parameters, but only performs directional order reduction preparation on the initial stiffness parameter portion related to the direction of insufficient stiffness of the support foot when the stiffness element of the support foot is lower than the set standing threshold. Since the stiffness attenuation compensation amount originates from the deviation characteristic value and is further combined with the path characteristic parameter, the stiffness attenuation compensation amount can transform the overall machine force distribution characteristics in the reaction force transmission relationship diagram into a correction amount that can directly act on the initial stiffness parameter. Through the stiffness attenuation compensation amount, when subsequently performing order reduction correction on the initial stiffness parameter, the stiffness requirement in the direction of insufficient stiffness in the support legs can be reduced while retaining the impedance foundation required for the operation of the dual-arm end arms, thereby reducing the adverse transmission of high stiffness at the dual-arm end arms to the four-legged support state.

[0057] Preferably, the specific implementation process of the step "using the stiffness attenuation compensation amount to perform order reduction correction on the initial stiffness parameter to generate a corrected impedance parameter, and determining the corrected impedance parameter as the target force potential impedance adjustment parameter for the ends of the two arms" is as follows: After the stiffness attenuation compensation amount has been formed, the initial stiffness parameter, the stiffness attenuation compensation amount, the direction field of the ends of the two arms, the current control cycle marker, and the support foot identifier are read first, and the initial stiffness parameter, the stiffness attenuation compensation amount, the direction field of the ends of the two arms, the current control cycle marker, and the support foot identifier are written into the impedance parameter correction record. The impedance parameter correction record is used to match the stiffness source of the initial stiffness parameter, the generation source of the stiffness attenuation compensation amount, and the direction field of the ends of the two arms that needs to be corrected, so that the order reduction correction process can be traced back to the insufficient stiffness of the support foot in the reaction force transmission relationship diagram. Subsequently, based on the direction field of the dual-arm ends, the stiffness component corresponding to the stiffness attenuation compensation amount is located in the initial stiffness parameters, and this stiffness component undergoes order reduction correction processing. The essence of this order reduction correction processing is to reduce the stiffness component in the corresponding direction according to the stiffness attenuation compensation amount without changing the stiffness source of the initial stiffness parameters, so that the impedance configuration of the dual-arm ends matches the support constraint capability of the supporting foot under the current posture. For stiffness components in the initial stiffness parameters not corresponding to the stiffness attenuation compensation amount, the original configuration of the stiffness component is retained to avoid irrelevant correction of the impedance configuration of the dual-arm ends in unrelated directions. After completing the order reduction correction processing, the corrected stiffness components, the uncorrected stiffness components, and the damping reference associated with the stiffness components are associated and encapsulated to generate the corrected impedance parameters. The technical essence of the modified impedance parameter is the resulting impedance configuration at the ends of the two arms, formed by introducing insufficient stiffness constraints on the supporting feet based on the initial stiffness parameter. The modified impedance parameter retains the stiffness source corresponding to the single-leg swing mark or full support mark, while also introducing constraints on the overall force distribution characteristics of the machine from the reaction force transmission relationship diagram. The modified impedance parameter is determined as the target force position impedance adjustment parameter for the ends of the two arms because it has completed continuous processing from the four-legged support state to the reaction force transmission relationship diagram and then to the stiffness attenuation compensation amount, enabling it to be used to subsequently generate posture compensation commands for driving the joints of the two arms according to the target force position impedance adjustment parameter. The technical essence of the target force position impedance adjustment parameter is the final impedance adjustment basis for the coordinated control of the force position at the ends of the two arms. It incorporates the support constraint capability of the supporting feet into the stiffness configuration of the ends of the two arms through order reduction correction processing, ensuring that the posture compensation commands at the ends of the two arms and the subsequent gait stabilization commands of the four-legged joints share a common force basis.

[0058] Optionally, the step of generating posture compensation commands for driving the biarm joints according to the target force potential impedance adjustment parameters, and generating gait stabilization commands for driving the quadrupedal joints after the posture compensation commands are issued, includes: Based on the end stiffness coefficient and end damping coefficient in the target force-position impedance adjustment parameters, determine the dynamic offset of the desired end contact force relative to the actual end contact force. The dynamic offset is converted into the expected joint torque value of the two arm joints; The posture compensation command is generated based on the expected value of the joint torque, and the posture compensation command is sent to the hardware controller of the two arm joints; After the attitude compensation command is issued, the target force impedance adjustment parameter is projected onto the joint space to determine the coupling reaction torque; The additional support force command for the four feet is generated based on the coupled reaction torque, and the gait stabilization command is generated based on the additional support force command.

[0059] Preferably, the specific implementation process of the step "determining the dynamic offset of the desired end contact force relative to the actual end contact force based on the end stiffness coefficient and end damping coefficient in the target force-position impedance adjustment parameters" is as follows: After the target force-position impedance adjustment parameters have been determined by the corrected impedance parameters, the end stiffness coefficient, end damping coefficient, dual-arm end direction field, and current control cycle flag in the target force-position impedance adjustment parameters are read first, and the desired end contact force given by the dual-arm end contact operation task in the current control cycle and the actual end contact force formed by the three-dimensional spatial force data in the dual-arm end operation parameters are read simultaneously to establish an end contact force offset calculation record. The end contact force offset calculation record is used to include the impedance source of the target force-position impedance adjustment parameters, the task source of the desired end contact force, and the sensing source of the actual end contact force in the same calculation caliber, so that the dynamic offset can be traced back to the force-position coordinated control process in the current control cycle. The technical essence of the desired end-effector contact force is the target contact force state that the ends of the two arms should maintain during the contact operation; the technical essence of the actual end-effector contact force is the real-time force state of the ends of the two arms at the current contact point, expressed by the three-dimensional spatial force data. Subsequently, the desired end-effector contact force and the actual end-effector contact force are decomposed according to the end-effector direction field to form contact force deviation components in each direction. These contact force deviation components express the force difference between the desired and actual end-effector contact forces under the same end-effector direction field. The direction decomposition process avoids mixing the force differences in the forward, lateral, and vertical directions into a single value that is difficult to use for joint control. Then, the contact force deviation components are subjected to stiffness mapping under positional resistance based on the end-effector stiffness coefficient to form a stiffness mapping result; and the actual end-effector contact force variation trend within a continuous control cycle is subjected to damping mapping based on the end-effector damping coefficient to form a damping mapping result. The stiffness mapping result is used to express the force adjustment requirement corresponding to the contact force deviation component under the constraint of the end stiffness coefficient, and the damping mapping result is used to express the change suppression requirement corresponding to the change trend of the actual end contact force under the constraint of the end damping coefficient. Subsequently, the stiffness mapping result and the damping mapping result are synthesized according to the same dual-arm end direction field to form the dynamic offset. The technical essence of the dynamic offset is to describe the difference between the desired end contact force and the actual end contact force under the impedance adjustment aperture; the dynamic offset reflects both the deviation of the magnitude and direction of the contact force in the current control cycle and the impact of the change trend of the actual end contact force on the stability of the dual-arm end contact operation in continuous control cycles.Through the aforementioned dynamic offset, the target force impedance adjustment parameters can be transformed from the end stiffness coefficient and end damping coefficient at the parameter level into the end force adjustment basis that can continue to enter the torque generation process of the dual-arm joint.

[0060] Preferably, in the specific technical implementation of the step "converting the dynamic offset into the expected joint torque value of the two arm joints", after the dynamic offset has been written into the end contact force offset calculation record, the dynamic offset, the joint rotation angle collected by the two arm joint encoder, the positive kinematic link, the end nodes of the two arms, and the trunk center of mass reference system are first read, and the dynamic offset, the joint rotation angle, the positive kinematic link, the end nodes of the two arms, and the trunk center of mass reference system are written into the expected joint torque value conversion record. The expected joint torque value conversion record is used to correspond the force space of the two arm end points where the dynamic offset is located with the joint space where the two arm joints are located, so that the dynamic offset will not stay at the end contact point of the two arms, but can be transmitted back to each of the two arm joints along the positive kinematic link. The technical essence of the force space at the end of the arms is the end-force adjustment space jointly expressed by the force direction, magnitude, and end-arm orientation fields of the end-arm nodes in the trunk center-of-mass reference system. The technical essence of the joint space is the control space jointly expressed by the joint rotation angles, joint arrangement order, and joint posture transformation relationships of each arm joint. In the joint space, each coordinate position corresponds to a rotational degree of freedom of an arm joint, and the value of each rotational degree of freedom is used to express the torque adjustment requirement that the arm joint should generate in the current control cycle. Subsequently, based on the posture transmission relationship between the end-arm nodes and each arm joint node in the forward kinematic link, the torque conversion relationship between the force space at the end of the arms and the joint space is determined. The torque conversion relationship does not directly assign the dynamic offset to the arm joints, but decomposes the dynamic offset to the rotational degrees of freedom of each arm joint according to the lever arm relationship, direction correspondence, and current joint rotation angle between the end-arm nodes and each arm joint node to form decomposed torque data of the arm joints. Next, the decomposed torque data of each bi-arm joint is superimposed in the same direction and canceled out in the opposite direction to form the expected joint torque value for the corresponding bi-arm joint. The technical essence of the expected joint torque value is the target torque data output by each bi-arm joint to bring the ends of the arms close to the desired end contact force. The expected joint torque value retains the directional information and impedance adjustment source in the dynamic offset and also corresponds to the joint rotation angle within the current control cycle. By converting the dynamic offset into the expected joint torque value, the contact force deviation at the ends of the arms can be transformed into an executable torque adjustment requirement for the bi-arm joints, providing a direct data basis for the subsequent generation of the attitude compensation command.

[0061] Preferably, the specific implementation process of the step "generating the attitude compensation command based on the expected joint torque value and sending the attitude compensation command to the hardware controller of the two arm joints" is as follows: After the expected joint torque value has been formed by the expected joint torque value conversion record, the expected joint torque value, the two arm joint identifiers, the current control cycle marker, the current angle of the two arm joints, and the torque output boundary of the two arm joints are read first, and the expected joint torque value, the two arm joint identifiers, the current control cycle marker, the current angle of the two arm joints, and the torque output boundary of the two arm joints are written into the attitude compensation command encapsulation record. The attitude compensation command encapsulation record is used to associate the expected joint torque value with the identifier of the two arm joints that need to perform torque adjustment, so that the subsequently generated attitude compensation command can be associated with each two arm joint. Subsequently, the expected joint torque value undergoes a torque output boundary verification process. This process compares the expected joint torque value with the torque output boundaries of the two-arm joints under the same torque caliber. If the expected joint torque value is within the torque output boundaries of the two-arm joints, it is retained and used as a scalable expected joint torque value. If the expected joint torque value exceeds the torque output boundaries of the two-arm joints, a boundary-limited expected joint torque value is generated based on these boundaries and used as a scalable expected joint torque value. This process avoids directly writing torque requirements exceeding the executable range of the two-arm joints into the attitude compensation instruction. Afterward, the scalable expected joint torque value is encapsulated according to the two-arm joint identifier, the current control cycle marker, and the instruction execution order to generate the attitude compensation instruction. The technical essence of the posture compensation command is a joint torque adjustment command for the two arm joints. It converts the end-effector force adjustment requirement derived from the target force-position impedance adjustment parameter into a posture compensation action source for the two arm joints within the current control cycle. When the posture compensation command is sent to the hardware controller of the two arm joints, the corresponding posture compensation command is written into the command receiving queue of the hardware controller according to the two arm joint identifier, and the current control cycle marker is used as the execution timing basis, enabling the hardware controller of the two arm joints to drive the corresponding two arm joints to perform torque adjustment according to the posture compensation command. By generating and sending the posture compensation command first, the force-position adjustment at the end of the two arms can first form a response basis at the two arm joint side, providing a time-defined data source for subsequently determining the coupling reaction torque based on the response results at the two arm joint side.

[0062] Preferably, the specific implementation process of the step "projecting the target force-position impedance adjustment parameter onto the joint space after the attitude compensation command is issued to determine the coupling reaction torque" is as follows: After the attitude compensation command has been written into the command receiving queue of the hardware controller of the two arm joints, the attitude compensation command, the target force-position impedance adjustment parameter, the expected value of the joint torque, the forward kinematic link, the reaction force transmission relationship graph, and the two arm joint identifiers are read first, and then the attitude compensation command, the target force-position impedance adjustment parameter, the expected value of the joint torque, the forward kinematic link, the reaction force transmission relationship graph, and the two arm joint identifiers are written into the coupling reaction torque calculation record. The coupling reaction torque calculation record is used to associate the attitude compensation command already issued by the two arm joints with the force transmission path from the ends of the two arms to the four legs in the reaction force transmission relationship graph, so that the coupling reaction torque can simultaneously reflect the attitude compensation action of the two arm joints and the force transmission relationship of the whole machine. Subsequently, a joint space is constructed based on the forward kinematics link and the joint identifiers of the two arms. This joint space is formed by the arrangement order of the two arm joints, the rotational degrees of freedom of each joint, the current joint rotation angle of each joint, and the attitude transformation relationship of the end-effectors of the two arms relative to the torso center of mass reference system. When projecting the target force potential impedance adjustment parameter onto the joint space, the end-effector stiffness coefficient and end-effector damping coefficient in the target force potential impedance adjustment parameter are first decomposed into the end-effector force space according to the end-effector direction field, forming directional end-effector stiffness and directional end-effector damping components. Then, based on the forward kinematics link, the directional end-effector stiffness and directional end-effector damping components are mapped onto the rotational degrees of freedom of each joint, forming joint space impedance projection data. The joint space impedance projection data is further processed in correspondence with the expected joint torque value to determine the contribution of the reverse torque transmitted by each joint to the robot torso when executing attitude compensation commands. Subsequently, combining the path feature parameters in the reaction force transmission relationship graph, the contribution of the reverse torque of each biarm joint is transferred and assigned along the graph connection relationship between the end nodes of the biarms, the torso centroid reference system, and the end nodes of the four legs, to form the coupled reaction torque. The technical essence of the coupled reaction torque is the description of the reverse torque transmitted from the robot to the four-legged support side through the robot torso when the biarm joints execute posture compensation commands. The coupled reaction torque is not a separate biarm joint torque, nor is it a separate force on the four legs, but rather the result of force transmission after the combined action of the posture compensation command, the target force impedance adjustment parameter, and the reaction force transmission relationship graph. Through the coupled reaction torque, additional support force commands can be generated on the four-legged end sides, ensuring that the gait stabilization commands of the four-legged joints and the posture compensation commands of the biarm joints maintain the same source of force.

[0063] Preferably, the specific implementation process of the step "generating additional support force commands for the four legs based on the coupled reaction torque, and generating the gait stabilization command based on the additional support force commands" is as follows: After the coupled reaction torque has been formed by the coupled reaction torque calculation record, the coupled reaction torque, the four-leg support phase parameter, the support period marker, the four-leg foot marker, the four-leg foot node in the reaction force transmission relationship diagram, and the ground contact force characteristics of each leg are read. The coupled reaction torque, the four-leg support phase parameter, the support period marker, the four-leg foot marker, the four-leg foot node, and the ground contact force characteristics of each leg are then written into the additional support force command generation record. The additional support force command generation record is used to associate the coupled reaction torque with the four legs in the support period during the current control cycle, so that the coupled reaction torque can be distributed to the four legs bearing ground support, and not to the four legs in the oscillation period. Subsequently, based on the support phase markers, the four-legged feet in the support phase are selected, and their corresponding identifiers are mapped to the four-legged foot nodes in the reaction force transmission relationship graph to form support foot allocation objects. Next, based on the path feature parameters and ground contact force characteristics of each leg in the reaction force transmission relationship graph, the coupled reaction torque is allocated to the support feet to form additional support force instructions corresponding to each support foot allocation object. The technical essence of these additional support force instructions is that the four-legged feet need to apply additional support force adjustment data to offset or share the coupled reaction torque; these additional support force instructions do not replace the original gait support force, but rather superimpose a support force adjustment amount corresponding to the biarm joint posture compensation action on the ground contact force characteristics expressed by each leg. This support force adjustment amount, as the core field of the additional support force instructions, continues to participate in the field encapsulation of subsequent gait stabilization instructions. Subsequently, the additional support force command, along with the corresponding quadrupedal foot end identifier, support period marker, current control cycle marker, and quadrupedal joint identifier, are encapsulated to generate the gait stabilization command. The technical essence of the gait stabilization command is a torque adjustment command oriented towards the quadrupedal joints. It converts the coupled reaction torque caused by the biarm joint posture compensation action into a support force adjustment requirement at the quadrupedal foot end, and further into stabilization control content that the quadrupedal joints can execute. Generating the gait stabilization command based on the additional support force command allows the quadrupedal joints to adjust the support force distribution after the biarm joints execute posture compensation commands, reducing the disturbance transmission caused by biarm end-contact operations on the quadrupedal support state. This enables the biarms and quadrupeds to perform coordinated force-position operations based on the same source of force within the same current control cycle.

[0064] Optionally, the step of generating additional support force commands at the ends of the four feet based on the coupled reaction torque, and generating the gait stabilization commands based on the additional support force commands, includes: Establish the overall machine reaction torque balance constraint conditions; Within the overall reaction torque balance constraint, the coupled reaction torque is distributed to each foot end in contact with the ground to generate additional support force commands for the four feet. Obtain the desired ground reaction force for each of the said feet corresponding to the additional support force command; Verify that the desired ground reaction force is within the anti-slip friction constraint boundary; If the desired ground reaction force is within the anti-slip friction constraint boundary, then a torque drive signal for each of the four legs is generated based on the desired ground reaction force. The gait stabilization command is generated based on the torque drive signal and sent to the hardware controller of the four-legged joints to prevent slippage of the four legs.

[0065] Preferably, the specific implementation process of the step "establishing the overall reaction torque balance constraint conditions" is as follows: After the coupled reaction torque has been formed by the coupled reaction torque calculation record, the coupled reaction torque, the four-leg support phase parameter, the support period mark, the swing period mark, the four-leg foot end mark, the four-leg foot end node in the reaction force transmission relationship diagram, the path characteristic parameter, and the ground contact force characteristics of each leg are read, and the coupled reaction torque, the four-leg support phase parameter, the support period mark, the swing period mark, the four-leg foot end mark, the four-leg foot end node, the path characteristic parameter, and the ground contact force characteristics of each leg are written into the overall reaction torque balance bearing record. The overall reaction torque balance load record is used to correlate the source of the reaction torque transmitted to the robot torso after the bi-arm joints execute the attitude compensation command, the source of the grounding status of each quadruped end within the current control cycle, and the path source of the quadruped end nodes in the reaction force transmission relationship graph. This ensures that the overall reaction torque balance constraint conditions originate from the joint constraints of the coupled reaction torque, the quadruped support phase parameters, and the reaction force transmission relationship graph, rather than static rules detached from the current control cycle. Subsequently, based on the support period marker, quadruped ends in contact with the ground are selected from the quadruped end identifiers, and node mapping is performed between the quadruped ends in contact with the ground and the quadruped end nodes in the reaction force transmission relationship graph to form a support foot balance object. Simultaneously, based on the swing period marker, quadruped ends not in contact with the ground are identified and excluded from the support foot balance object. The supporting leg balance object represents the four-legged foot-end nodes capable of absorbing the coupled reaction torque within the current control cycle. This supporting leg balance object subsequently participates in the distribution of the coupled reaction torque. Next, based on the path characteristic parameters, the torque transmission share of the coupled reaction torque reaching each supporting leg balance object is determined. Then, based on the ground contact force characteristics of each leg, the existing grounding foundation of each supporting leg balance object is determined. Finally, the torque transmission share and the grounding foundation are jointly written into the overall machine reaction torque balance bearing record. The established overall machine reaction torque balance constraint condition limits the distribution of the coupled reaction torque to the four-legged foot-ends in contact with the ground, and requires that the distributed additional support force command and the ground contact force characteristics of each leg together form a counteracting relationship against the reverse torque of the robot's torso. The technical essence of the overall machine counter-torque balance constraint condition is to convert the counter-torque introduced by the attitude compensation of the two arms into a support force distribution constraint that can be shared by the four legs, so that when generating the additional support force command at the end of the four legs, the number of supporting foot balance objects, the grounded foundation and the torque transmission share can be considered at the same time.

[0066] Preferably, in the specific technical implementation of the step "distributing the coupled reaction torque to each foot end in contact with the ground within the overall machine reaction torque balance constraint condition to generate the additional support force command for the four feet", after the overall machine reaction torque balance constraint condition has been written into the overall machine reaction torque balance load record, the coupled reaction torque, the supporting foot balance object, the path characteristic parameter, the torque transmission share, and the ground contact force characteristics of each leg in the overall machine reaction torque balance load record are first read, and the coupled reaction torque, the supporting foot balance object, the path characteristic parameter, the torque transmission share, and the ground contact force characteristics of each leg are written into the additional support force allocation record. The additional support force allocation record is used to correspond the allocation relationship between each supporting foot balance object and the coupled reaction torque, so that the coupled reaction torque can be decomposed according to the grounding state and force transmission path of the four foot end nodes. Subsequently, the coupled reaction torque is path-assigned based on the path characteristic parameters to correct the torque transmission share of the coupled reaction torque reaching each supporting foot balance object along different pattern connections. This torque transmission share expresses the proportion of the reverse torque that each supporting foot balance object should bear, and it continues to participate in subsequent grounding force correction processing. Next, grounding force correction processing is performed on the torque transmission share based on the grounding contact force characteristics of each leg, ensuring that supporting foot balance objects with higher grounding force foundations bear more additional support force adjustment, and supporting foot balance objects with lower grounding force foundations bear less additional support force adjustment. This grounding force correction processing does not directly replace the grounding contact force characteristics of each leg with additional support force commands; instead, it uses the grounding contact force characteristics of each leg as the existing support foundation to adjust the torque transmission share of the coupled reaction torque. After completing the grounding force correction process, the additional support force adjustment amount corresponding to each supporting foot balance object is converted into the foot end support force direction and magnitude. This is then combined with the four-leg foot end identifier, the current control cycle marker, and the support period marker to encapsulate the data and generate the additional support force command for each of the four legs. The technical essence of this additional support force command is to add support force adjustment data to the four legs in contact with the ground, based on the grounding force characteristics expressed by the original leg grounding contact force features. This additional support force command is used to share the coupling reaction torque generated after the biarm joints execute posture compensation commands, enabling subsequent gait stabilization commands to use the additional support force adjustment amount on the four-leg foot end side as input, rather than generating four-leg joint control content solely based on the original gait state.

[0067] Preferably, the specific implementation process of step "obtaining the expected ground reaction force corresponding to the additional support force command for each of the four legs" is as follows: After the additional support force command for each of the four legs has been generated by the additional support force allocation record, the additional support force command, the ground contact force characteristics of each leg, the identification of each of the four legs, the support period marker, and the current control cycle marker are read first, and then the additional support force command, the ground contact force characteristics of each leg, the identification of each of the four legs, the support period marker, and the current control cycle marker are written into the expected ground reaction force calculation record. The expected ground reaction force calculation record is used to make the existing grounding force foundation of each of the four legs correspond to the additional support force command corresponding to that four legs, so that the expected ground reaction force can simultaneously reflect the adjustment amount of the additional support force introduced by the original support force and the side reaction force of the two arms. Subsequently, based on the quadrupedal foot markers, the ground contact force characteristics of each leg corresponding to each quadrupedal foot in the support phase are used as the ground reaction force basis for that quadrupedal foot, and the additional support force command corresponding to that quadrupedal foot is used as the ground reaction force adjustment amount for that quadrupedal foot. The ground reaction force basis is used to express the support force state that the quadrupedal foot has formed with the ground in the current control cycle, and the ground reaction force adjustment amount is used to express the support force state that needs to be added or adjusted at the quadrupedal foot to share the coupled reaction torque. Next, under the same foot coordinate caliber, the ground reaction force basis and the ground reaction force adjustment amount are checked for directional consistency; when the ground reaction force basis and the ground reaction force adjustment amount have the same directional component, the same directional component is superimposed; when the ground reaction force basis and the ground reaction force adjustment amount have opposite directional components, the opposite directional components are canceled out to form the desired ground reaction force corresponding to that quadrupedal foot. The technical essence of the desired ground reaction force is the ground support force result that the four legs should achieve after executing the additional support force command. The desired ground reaction force retains both the existing ground support state expressed by the ground contact force characteristics of each leg and the reaction torque sharing requirement expressed by the additional support force command. By obtaining the desired ground reaction force of each of the four legs corresponding to the additional support force command, the anti-slip friction constraint boundary of the desired ground reaction force can be checked, thereby avoiding the four legs from forming a support force requirement that exceeds the friction constraint capacity of the leg ends in order to share the coupling reaction torque.

[0068] Preferably, the specific implementation process of the step "verifying whether the expected ground reaction force is within the anti-slip friction constraint boundary" is as follows: After the expected ground reaction force has been written into the expected ground reaction force calculation record, the expected ground reaction force, the four-legged foot end markers, the support period marker, the ground contact force characteristics of each leg, and the foot end contact surface state data are read first, and then written into the anti-slip friction constraint verification record. The foot end contact surface state data comes from the contact pressure change, the foot end tangential slight slip trend, and the foot end contact normal direction when the four-legged feet contact the ground, and is used to express the contact state between the four-legged feet and the ground within the current control cycle. Subsequently, the foot end normal support capacity and foot end tangential friction capacity are determined based on the foot end contact surface state data, and the anti-slip friction constraint boundary is formed based on the foot end normal support capacity and foot end tangential friction capacity. The technical essence of the anti-slip friction constraint boundary is to define the data boundaries of the tangential and normal support force ranges that the four legs can withstand when in contact with the ground. The anti-slip friction constraint boundary is not a fixed value set separately, but rather formed jointly based on the foot contact surface state data, the ground contact force characteristics of each leg, and the support period marker within the current control cycle. Next, the desired ground reaction force is decomposed according to the foot's normal and tangential directions to obtain the desired normal reaction force component and the desired tangential reaction force component. The desired normal reaction force component expresses the degree of vertical support that the four legs need from the ground, and the desired tangential reaction force component expresses the degree of horizontal anti-slip resistance that the four legs need from the ground. Subsequently, the desired normal reaction force component is compared in the same direction with the foot's normal support capacity, and the desired tangential reaction force component is compared in the same direction with the foot's tangential friction capacity to determine whether the desired ground reaction force is within the anti-slip friction constraint boundary. When both the desired normal reaction force component and the desired tangential reaction force component are within the anti-slip friction constraint boundary, it is determined that the desired ground reaction force is within the anti-slip friction constraint boundary. This verification process is used to eliminate support force requirements that could cause slippage at the feet before generating the torque drive signal for the quadrupedal joints, ensuring that the subsequent torque drive signal corresponds to the range of support force that the contact surface at the feet can withstand.

[0069] Preferably, in the specific technical implementation of the step "if the desired ground reaction force is within the anti-slip friction constraint boundary, then generate torque drive signals for each quadruped joint based on the desired ground reaction force", after the anti-slip friction constraint verification record indicates that the desired ground reaction force is within the anti-slip friction constraint boundary, the desired ground reaction force, the quadruped foot end identifier, the stance phase marker, the inverse kinematic link, the leg joint posture transformation relationship, and the quadruped joint identifier are read first, and then written into the torque drive signal generation record. The torque drive signal generation record is used to associate the desired ground reaction force at the quadruped foot end level with the torque output object at the quadruped joint level, so that the desired ground reaction force can be converted into a torque drive signal executable by the quadruped joints along the inverse kinematic link. Subsequently, the leg joint node arrangement order corresponding to the four-legged foot end during the support phase is determined based on the foot end identifiers, and the contribution relationship of each four-legged joint to the support force direction of the four-legged foot end is determined based on the leg joint posture transformation relationship. This contribution relationship expresses the degree of influence of a certain four-legged joint's rotational degree of freedom on the desired normal reaction force component and the desired tangential reaction force component. Next, the desired normal reaction force component and the desired tangential reaction force component in the desired ground reaction force are decomposed along the inverse kinematic link to each four-legged joint in the corresponding leg joint node arrangement order to form four-legged joint decomposed torque data. The four-legged joint decomposed torque data is further processed according to the four-legged joint identifiers and boundary verification is performed according to the four-legged joint torque output boundary to form torque drive signals for each four-legged joint. The technical essence of the torque-driven signal is to convert the desired ground reaction force at the ends of the four feet into target torque data that each of the four foot joints needs to output within the current control cycle. This torque-driven signal is not generated solely based on gait rhythm, but rather originates simultaneously from the coupled reaction torque, the additional support force command, the desired ground reaction force, and the anti-slip friction constraint boundary verification result. Through this torque-driven signal, subsequent gait stabilization commands can incorporate the torque sharing requirements caused by the biarm joint posture compensation movements into the torque output of the four foot joints, ensuring that the support force adjustment at the ends of the four feet and the posture compensation movements of the biarm joints maintain a common force relationship.

[0070] Preferably, the specific implementation process of the step "generating the gait stabilization command based on the torque drive signal and issuing it to the hardware controller of the quadrupedal joints to prevent slippage of the quadrupedal joints through the hardware controller" is as follows: After the torque drive signal has been generated and recorded, the torque drive signal, the quadrupedal joint identifier, the quadrupedal foot end identifier, the current control cycle marker, the support phase marker, and the anti-slip friction constraint verification record are read first, and then written into the gait stabilization command encapsulation record. The gait stabilization command encapsulation record is used to correlate the torque drive signal at the quadrupedal joint level, the support phase state at the quadrupedal foot end level, and the anti-slip friction constraint verification result, so that the gait stabilization command can express the stabilization torque output requirement of the quadrupedal joints within the current control cycle. Subsequently, the torque drive signals are encapsulated by joint affixation based on the quadrupedal joint identifiers, and the torque drive signals of multiple quadrupedal joints corresponding to the same quadrupedal foot end are configured into the same foot end support adjustment unit based on the quadrupedal foot end identifiers. The foot end support adjustment unit is used to express the coordinated torque output relationship of multiple quadrupedal joints required for a quadrupedal foot end in the support phase to achieve the desired ground reaction force. Next, the foot end support adjustment unit, the current control cycle marker, and the anti-slip friction constraint verification record are encapsulated together into the gait stabilization command. The technical essence of the gait stabilization command is a stabilization torque control command for the quadrupedal joints under the condition that the desired ground reaction force is already within the anti-slip friction constraint boundary; the gait stabilization command incorporates the foot end anti-slip constraint, the additional support force command, and the quadrupedal joint torque drive signals into the same execution caliber. When the gait stabilization command is sent to the hardware controller of the quadrupedal joints, the corresponding gait stabilization command is written into the command receiving queue of the hardware controller of the quadrupedal joints according to the quadrupedal joint identifier, and the current control cycle flag is used as the execution timing basis, so that the hardware controller of the quadrupedal joints drives the corresponding quadrupedal joints to output torque according to the gait stabilization command. Since the gait stabilization command has been checked by the anti-slip friction constraint boundary before generation, when the hardware controller of the quadrupedal joints executes the gait stabilization command, it causes the feet of the quadrupedal joints in the support phase to adjust the normal support and tangential support according to the expected ground reaction force, thereby suppressing slippage of the feet and allowing the coupled reaction torque caused by the biarm joint posture compensation action to be distributed on the quadrupedal support side.

[0071] Optionally, obtaining the quadrupedal support phase parameters and the end-effector manipulation parameters of the centaur robot includes: Contact force sensors deployed at the ends of the four feet collect the plantar contact pressure values; The foot end where the sole contact pressure value continuously exceeds the contact threshold is marked as being in the support phase; The foot end where the sole contact pressure value does not exceed the contact threshold is marked as being in the swing phase; The quadrupedal support phase parameters are constructed based on the support phase markers and swing phase markers corresponding to each foot end; Three-dimensional force data are collected by a six-axis torque sensor deployed at the ends of the twin arms; The joint rotation angle is acquired using a dual-arm joint encoder; The three-dimensional spatial force data and the joint rotation angle are integrated and processed to generate the end-effector operation parameters of the two arms.

[0072] Preferably, the specific implementation process of the step "collecting plantar contact pressure values ​​through contact force sensors deployed at the ends of the four legs" is as follows: After the Centaur robot enters the current control cycle, the four-legged end acquisition configuration record is first read. The four-legged end acquisition configuration record is established when the Centaur robot completes assembly calibration and basic control parameter configuration. The four-legged end acquisition configuration record includes the four-legged end identifier, contact force sensor identifier, contact force sensor installation position, contact force sensor sampling direction, plantar contact direction field, current control cycle marker, data arrangement caliber of plantar contact pressure values, and reading method of contact threshold configuration record. The four-legged end acquisition configuration record is used to associate each four-legged end with the contact force sensor deployed at that four-legged end, and to associate the contact force sensor sampling direction with the plantar contact direction through the plantar contact direction field, so that the plantar contact pressure value can be traced back to the specific four-legged end identifier, contact force sensor identifier, and plantar contact direction field. Subsequently, based on the current control cycle marker, each contact force sensor is triggered to perform synchronous sampling, and the contact pressure sample value corresponding to the foot contact direction field is read from the sampling output of each contact force sensor. After the contact pressure sample value is entered into the plantar pressure processing record, the contact pressure sample values ​​of the same four feet end are first processed for location assignment based on the installation position of the contact force sensor to form location-assigned pressure sample values; then, the location-assigned pressure sample values ​​are processed for direction normalization based on the sampling direction of the contact force sensor and the foot contact direction field to form the plantar contact pressure value corresponding to each of the four feet end. The plantar pressure processing record is used to record the processing source relationship between the contact pressure sample value, the location-assigned pressure sample value, and the plantar contact pressure value, so that the plantar contact pressure value is not independent data detached from the sampling output. The technical essence of the plantar contact pressure value is that it is grounding force data collected by contact force sensors when the four feet contact the ground, and then processed by position attribution and direction normalization. This plantar contact pressure value is not a single instantaneous sensor output, but rather a grounding status judgment criterion established by corresponding with the four-foot identification, the current control cycle marker, and the plantar contact direction field. Through this plantar contact pressure value, it is possible to subsequently determine whether each of the four feet is in the support or swing phase, and further construct the four-foot support phase parameters, so that the identification of the four-foot support state comes from the actual contact data of the four-foot ends, rather than solely based on a pre-set gait rhythm.

[0073] Preferably, in the specific technical implementation of the step "marking the foot tip of the quadruped whose plantar contact pressure value continuously exceeds the contact threshold as being in the support phase", after the plantar contact pressure value has been written into the plantar pressure compilation record, the contact threshold configuration record is first read according to the reading method of the contact threshold configuration record in the quadruped foot acquisition configuration record, and the plantar contact pressure value, the quadruped foot identifier, the current control cycle mark, and the plantar contact direction field in the plantar pressure compilation record are read simultaneously. Then, the plantar contact pressure value, the quadruped foot identifier, the current control cycle mark, the plantar contact direction field, and the contact threshold configuration record are written into the support phase mark determination record. The contact threshold configuration record is established when configuring the basic control parameters. The contact threshold configuration record is configured according to the no-load offset of the contact force sensor, the minimum contact pressure range of the quadruped foot tip under normal support, and the sampling fluctuation range under ground contact conditions. The contact threshold configuration record includes the contact threshold, the continuous exceedance determination duration, the quadruped foot identifier, and the plantar contact direction field to which the contact threshold applies. The technical essence of the contact threshold is the pressure threshold that distinguishes whether the four-legged foot end forms effective ground support. This contact threshold is not a single empirical value, but rather is configured in conjunction with the sampling noise of the contact force sensor, the minimum ground support requirement of the four-legged foot end, the foot contact direction field, and the sampling cycle of the current control period. Subsequently, the foot contact pressure value of the same four-legged foot end within a continuous control cycle is compared with the contact threshold under the same pressure caliber to form a support period pressure comparison result. This support period pressure comparison result is continuously checked based on the continuous exceedance determination time to verify whether the foot contact pressure value continuously exceeds the contact threshold. When the support period pressure comparison result for a certain four-legged foot end shows that the foot contact pressure value exceeds the contact threshold within the continuous exceedance determination time, the foot end support status field of that four-legged foot end is written as a support period marker. The technical essence of the support period marker is the data representation of the effective grounding support state of the four-legged feet. The support period marker is subsequently involved in the construction of the four-legged support phase parameters and serves as the data source for identifying single-leg swing markers, full support markers, and filtering the four-legged feet in contact with the ground when generating gait stabilization commands.

[0074] Preferably, the specific implementation process of the step "marking the foot tip of the quadruped that the sole contact pressure value does not exceed the contact threshold as being in the swing phase" is as follows: After the support phase marking determination record has completed the comparison processing of the sole contact pressure value and the contact threshold, the sole contact pressure value, the contact threshold, the quadruped foot tip identifier, the current control cycle marker, the sole contact direction field, and the continuous non-exceeding determination duration in the support phase marking determination record are read again, and the sole contact pressure value, the contact threshold, the quadruped foot tip identifier, the current control cycle marker, the sole contact direction field, and the continuous non-exceeding determination duration are written into the swing phase marking determination record. The continuous non-exceeding determination duration and the contact threshold are written together into the contact threshold configuration record to distinguish between short-term contact pressure fluctuations of the quadruped foot tip and the state where the quadruped foot tip is truly detached from the ground support, avoiding frequent switching between the support phase marking and the swing phase marking caused by instantaneous sampling jitter. Subsequently, the plantar contact pressure value of the same quadruped foot end within a continuous control cycle is compared with the contact threshold under the same pressure caliber to form a swing period pressure comparison result. This swing period pressure comparison result is continuously checked based on the continuous "not exceeding the judgment time" to verify whether the plantar contact pressure value does not exceed the contact threshold. When the swing period pressure comparison result for a certain quadruped foot end shows that the plantar contact pressure value does not exceed the contact threshold within the continuous "not exceeding the judgment time," the foot end support state field of that quadruped foot end is written as a swing period marker. The technical essence of the swing period marker is a data representation of the quadruped foot end not forming an effective grounding support state; the swing period marker does not simply indicate that the quadruped foot end is raised, but rather that the quadruped foot end cannot serve as a grounding support object to receive the coupling reaction torque within the current control cycle. By using the swing period marker, the four-legged foot ends in the support period and the four-legged foot ends in the swing period can be distinguished when constructing the four-legged support phase parameters. The four-legged foot ends in the swing period are excluded when generating the overall machine reaction torque balance constraint conditions. This ensures that the four-legged support state identification and the force distribution process in the force-position coordinated control maintain the same data source.

[0075] Preferably, the specific implementation process of the step "constructing the quadrupedal support phase parameter according to the support period marker and swing period marker corresponding to each of the quadrupedal foot ends" is as follows: After the support period marker and the swing period marker are formed by the support period marker determination record and the swing period marker determination record, respectively, the quadrupedal foot end identifier, support period marker, swing period marker, current control cycle marker, and foot end support state field corresponding to each quadrupedal foot end are read first, and the quadrupedal foot end identifier, the support period marker, the swing period marker, the current control cycle marker, and the foot end support state field are written into the quadrupedal support phase parameter construction record. The quadrupedal support phase parameter construction record is used to incorporate the support period markers and swing period markers scattered under different quadrupedal foot end identifiers into the same control cycle caliber, so that the quadrupedal support phase parameter can express the quadrupedal support configuration of the whole machine, rather than just expressing the contact state of a single quadrupedal foot end. Subsequently, according to the installation order of the quadrupedal foot ends in the Centaur robot, the support period marker and swing period marker corresponding to each quadrupedal foot end are phase-arranged to form a quadrupedal foot end support phase sequence. Each position in the quadrupedal foot support phase sequence corresponds to a quadrupedal foot identifier, and each position records the support period marker or swing period marker corresponding to that quadrupedal foot identifier within the current control cycle. Next, based on the quadrupedal foot support phase sequence, the number of quadrupedal feet in the support period and the number of quadrupedal feet in the swing period are merged to form a quadrupedal foot support quantity merging result. This result is then written into the quadrupedal support phase parameter construction record to form the quadrupedal support phase parameter. The technical essence of the quadrupedal support phase parameter is that it is a whole-machine support state data composed of the support period marker, swing period marker, quadrupedal foot identifier, current control cycle marker, and quadrupedal foot support quantity merging result corresponding to each quadrupedal foot. The quadrupedal support phase parameter subsequently participates in the mapping of the closed-loop kinematic chain, the parsing of single-leg swing markers or full support markers, the configuration of initial stiffness parameters, and the generation of gait stabilization commands. Through the four-legged support phase parameters, the contact state of the four-legged feet can be converted into a readable, indexable, and transferable support phase basis for subsequent force-position coordinated control.

[0076] Preferably, the specific implementation process of the step "collecting three-dimensional spatial force data by deploying a six-axis torque sensor at the end of the two arms" is as follows: After the Centaur robot enters the current control cycle, the end-of-arm acquisition configuration record is read. This record is established when the Centaur robot completes assembly calibration and basic control parameter configuration. The end-of-arm acquisition configuration record includes end-of-arm identifiers, a six-axis torque sensor identifier, the installation position of the six-axis torque sensor, the sampling direction of the six-axis torque sensor, an end-of-arm direction field, and a current control cycle marker. The end-of-arm acquisition configuration record is used to associate the end of the two arms with the six-axis torque sensor deployed at the end of the two arms, so that the subsequently generated three-dimensional spatial force data can be traced back to the specific end-of-arm identifier and the six-axis torque sensor identifier. Subsequently, the six-axis torque sensor is triggered to sample according to the current control cycle marker, and the force sampling values ​​of the end of the two arms in the forward, lateral, and vertical directions are read from the sampling output of the six-axis torque sensor. After the force sampling values ​​are entered into the three-dimensional space force data generation record, the force sampling values ​​are first processed to assign the action position according to the installation position of the six-axis torque sensor to form the action position assigned force data. Then, the action position assigned force data is processed to convert the direction according to the sampling direction of the six-axis torque sensor and the direction field of the dual-arm end effector to form the three-dimensional space force data. The three-dimensional space force data generation record is used to record the processing source relationship between the force sampling values, the action position assigned force data, and the three-dimensional space force data, so that the three-dimensional space force data can be traced back to the sampling output of the six-axis torque sensor. The technical essence of the three-dimensional space force data is the force data formed under a unified spatial reference when the dual-arm end effector is subjected to external force at the current contact operation position. The three-dimensional space force data is not an isolated sensor output, but an end effector force description after establishing a correspondence with the dual-arm end effector identifier, the dual-arm end effector direction field, and the current control cycle marker. The three-dimensional spatial force data can be used to generate the end-effector operation parameters of the dual arms. When determining the dynamic offset of the desired end-effector contact force relative to the actual end-effector contact force, these parameters can be used as the sensing source of the actual end-effector contact force, enabling the contact operation state of the dual arms to participate in the coordinated generation of attitude compensation commands and gait stabilization commands.

[0077] Preferably, in the specific technical implementation of the step "collecting joint rotation angles through dual-arm joint encoders", after the three-dimensional spatial force data has been written into the three-dimensional spatial force data generation record, the dual-arm joint encoder configuration record is read. This dual-arm joint encoder configuration record is established when the dual-arm mechanism completes assembly calibration. The dual-arm joint encoder configuration record includes dual-arm joint identifiers, dual-arm joint encoder identifiers, dual-arm joint encoder zero-position calibration data, dual-arm joint rotation direction field, dual-arm joint arrangement order, and current control cycle marker. The dual-arm joint encoder configuration record is used to establish a correspondence between each dual-arm joint and its corresponding dual-arm joint encoder, enabling the joint rotation angle to be traced back to the specific dual-arm joint identifier and dual-arm joint encoder identifier. Subsequently, based on the current control cycle marker, each dual-arm joint encoder is triggered to read its current encoded value, and the current encoded value is zero-position corrected based on the dual-arm joint encoder zero-position calibration data to form a zero-position corrected encoded value. Then, based on the dual-arm joint rotation direction field, the zero-position corrected encoded value is directionally assigned to form the joint rotation angle corresponding to each dual-arm joint. The dual-arm joint encoder configuration record continues to record the processing source relationship between the current encoded value, the zero-position correction encoded value, and the joint rotation angle, enabling the joint rotation angle to be traced back to the corresponding dual-arm joint encoder identifier. The technical essence of the joint rotation angle is the angular state data of the dual-arm joints relative to the assembly calibration zero position within the current control cycle. The joint rotation angle, together with the arrangement order of the dual-arm joints, expresses the current posture of the dual-arm mechanism, enabling the subsequent establishment of a positive kinematic link from the end-effectors of the arms to the torso's center of mass reference system, and converting the dynamic offset in the force space of the end-effectors into the expected joint torque values ​​of the dual-arm joints. Through the joint rotation angle, the three-dimensional spatial force data no longer remains at the level of force description at the end-effectors of the arms, but can correspond to the posture of the dual-arm joints, providing a joint posture source for the subsequent generation of the operational parameters of the dual-arm end-effectors.

[0078] Preferably, the specific implementation process of the step "integrating the three-dimensional spatial force data and the joint rotation angle to generate the dual-arm end-effector operation parameters" is as follows: After the three-dimensional spatial force data and the joint rotation angle are generated from the three-dimensional spatial force data generation record and the dual-arm joint encoder configuration record, respectively, the three-dimensional spatial force data, the joint rotation angle, the dual-arm end-effector identifier, the dual-arm joint identifier, the dual-arm joint arrangement order, the dual-arm end-effector direction field, and the current control cycle marker are read first, and then written into the dual-arm end-effector operation parameter generation record. The dual-arm end-effector operation parameter generation record is used to incorporate the force state of the dual-arm end-effectors and the posture state of the dual-arm joints into the same control cycle, so that the dual-arm end-effector operation parameters can simultaneously express the three-dimensional spatial action on the dual-arm end-effectors and the current angle posture of the dual-arm joints. Subsequently, based on the end-effector identifiers of both arms, the three-dimensional spatial force data is associated with the corresponding end-effectors for end-effector attribution processing. Then, based on the joint identifiers and the joint arrangement order of both arms, the joint rotation angle is associated with the corresponding joint to form a correspondence between end-effector force and joint posture. This correspondence is further processed using the end-effector direction field to ensure that the force direction in the three-dimensional spatial force data corresponds to the arm posture direction expressed by the joint rotation angle under a unified spatial reference. Afterward, the end-effector force and joint posture correspondence is written into the end-effector operation parameter generation record to generate the end-effector operation parameters. The technical essence of the end-effector operation parameters is that they are formed by the three-dimensional spatial force data, the joint rotation angle, the end-effector identifiers, the joint identifiers, the joint arrangement order, the end-effector direction field, and the current control cycle marker, creating end-effector operation state data. These end-effector operation parameters subsequently participate in the mapping of the closed-loop kinematic chain, the determination of dynamic offsets, the conversion of expected joint torque values, and the generation of posture compensation commands. Through the aforementioned end-effector operation parameters, the contact operation force state of the end-effectors and the posture state of the arm joints can be integrated into the same data entry point, providing a transferable end-effector operation basis for the force-position coordination control of the four legs and the arms.

[0079] like Figure 3 This application provides an embodiment of a force-position coordination control device for the quadrupedal and dual-arm systems of a centaur robot, comprising: The parameter acquisition module is configured to acquire the quadrupedal support phase parameters and the end-effector operation parameters of the Centaur robot. The closed-loop kinematic chain mapping module is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The relationship graph construction module is configured to construct a reaction force transmission relationship graph that characterizes the force distribution characteristics of the whole machine based on the path feature parameters that characterize the transmission of contact force at the ends of the two arms to the four legs in the closed-loop kinematic chain. The initial stiffness configuration module is configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms according to the single-leg swing mark or full-support mark. The impedance adjustment generation module is configured to compensate the initial stiffness parameter based on the reaction force transmission relationship spectrum to generate a target force potential impedance adjustment parameter for the ends of the double arms. The collaborative instruction generation module is configured to generate posture compensation instructions for driving the joints of both arms according to the target force-position impedance adjustment parameters, and to generate gait stabilization instructions for driving the joints of the four legs after the posture compensation instructions are issued, so that the arms and the four legs perform force-position collaborative operations.

[0080] like Figure 4 As shown, an electronic device according to an embodiment of this application includes a memory and a processor; The memory stores computer programs; When the processor executes the computer program, it implements any of the methods described in this application.

[0081] like Figure 5 As shown, this is a computer-readable storage medium according to an embodiment of the present application, wherein the computer-readable storage medium stores a computer program; When the computer program is executed by a processor, it implements any of the methods described in this application.

[0082] like Figure 6 As shown, this is an embodiment of a centaur robot according to this application, which includes a robot torso, a quadruped mechanism, a dual-arm mechanism, a sensor acquisition component, a dual-arm joint hardware controller, a quadruped joint hardware controller, and a control processor. The quadruped mechanism is connected to the robot's torso, and the quadruped mechanism includes four leg joints and four leg ends; The dual-arm mechanism is connected to the robot's torso, and the dual-arm mechanism includes dual-arm joints and dual-arm ends; The sensing and acquisition component is configured to acquire the four-leg support phase parameters corresponding to the four-legged mechanism and the end-effector operation parameters corresponding to the dual-arm mechanism. The control processor is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The control processor is also configured to construct a reaction force transmission relationship map representing the force distribution characteristics of the whole machine based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs. The control processor is also configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms according to the single-leg swing mark or full-support mark. The control processor is also configured to compensate the initial stiffness parameter based on the reaction force transmission relationship diagram to generate a target force potential impedance adjustment parameter for the ends of the two arms. The control processor is also configured to generate attitude compensation commands for driving the bi-arm joints according to the target force impedance adjustment parameters, and to send the attitude compensation commands to the bi-arm joint hardware controller. The control processor is also configured to generate gait stabilization commands for driving the quadrupedal joints after the attitude compensation command is issued, and to issue the gait stabilization commands to the quadrupedal joint hardware controller so that the dual-arm mechanism and the quadrupedal mechanism perform force-position coordinated operations.

[0083] like Figure 7 As shown, this is an embodiment of the force-position coordination control system for a centaur robot with four legs and two arms, which includes a centaur robot, a parameter acquisition unit, a coordination control unit, a hardware controller for the two arm joints, and a hardware controller for the four leg joints. The centaur robot includes a quadruped mechanism and a dual-arm mechanism; The parameter acquisition unit is configured to acquire the four-legged support phase parameters corresponding to the four-legged mechanism and the double-arm end operation parameters corresponding to the double-arm mechanism. The collaborative control unit is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The collaborative control unit is also configured to construct a reaction force transmission relationship map representing the force distribution characteristics of the whole machine based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs. The collaborative control unit is also configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms of the dual-arm mechanism according to the single-leg swing mark or the full-support mark. The collaborative control unit is also configured to compensate the initial stiffness parameter based on the reaction force transmission relationship diagram to generate a target force potential impedance adjustment parameter for the ends of the two arms. The collaborative control unit is also configured to generate attitude compensation commands for driving the dual-arm joints of the dual-arm mechanism according to the target force impedance adjustment parameters, and send the attitude compensation commands to the dual-arm joint hardware controller. The collaborative control unit is also configured to generate gait stabilization commands for driving the quadrupedal joints of the quadrupedal mechanism after the attitude compensation command is issued, and send the gait stabilization commands to the quadrupedal joint hardware controller so that the dual-arm mechanism and the quadrupedal mechanism perform force-position collaborative operations.

[0084] Figures 2-7 For an exemplary description, please refer to the above. Figure 1 This will not be elaborated upon here.

Claims

1. A method for force-position coordinated control of the quadrupeds and arms of a centaur robot, characterized in that, include: Obtain the quadrupedal support phase parameters and the end-effector operation parameters of the centaur robot, and map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. Based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs, a reaction force transmission relationship diagram characterizing the force distribution characteristics of the whole machine is constructed. The single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter is analyzed to identify the quadrupedal support state of the centaur robot, and the initial stiffness parameter is configured for the end of the two arms based on the single-leg swing mark or full-support mark. The initial stiffness parameter is compensated based on the reaction force transmission relationship diagram to generate the target force potential impedance adjustment parameter for the ends of the double arms. According to the target force-position impedance adjustment parameters, posture compensation commands for driving the joints of both arms are generated, and after the posture compensation commands are issued, gait stabilization commands for driving the joints of the four legs are generated, so that the arms and the four legs perform force-position coordinated operations.

2. The method for force-position coordinated control of the quadrupeds and arms of a centaur robot according to claim 1, characterized in that, The process of mapping the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain includes: Obtain the trunk centroid reference system and use the trunk centroid reference system as a unified spatial reference for the quadrupedal support phase parameter and the end-effector operation parameter of the two arms. Based on the operational parameters of the two arms, a positive kinematic link is established from the ends of the two arms to the reference system of the trunk center of mass; Based on the four-legged support phase parameters, an inverse kinematic link is established from each foot end of the four legs to the trunk center of mass reference system; Using the trunk centroid reference system as the splicing reference, the forward kinematic links and the reverse kinematic links are spliced ​​together to generate a global spatial topology structure that runs through the arms, trunk and four legs. Extract the ground contact force characteristics of each leg included in the four-legged support phase parameters; Extract the joint torque characteristics included in the end-effector operation parameters of the two arms; The ground contact force characteristics of each leg and the joint torque characteristics are configured into the corresponding nodes of the global spatial topology to generate the closed-loop kinematic chain.

3. The method for force-position coordinated control of the quadrupeds and arms of a centaur robot according to claim 1, characterized in that, The step of constructing a reaction force transmission relationship map characterizing the force distribution characteristics of the entire machine based on the path characteristic parameters representing the transmission of contact force at the ends of the two arms to the four legs in the closed-loop kinematic chain includes: Read the Jacobian transfer matrix elements in the closed-loop kinematic chain; Based on the Jacobian transfer matrix elements, the six-degree-of-freedom trunk perturbation vector caused by the interaction force at the ends of the arms is identified. Along the force transmission path of each leg joint of the four legs, the attenuation feature of the six-degree-of-freedom perturbation vector of the trunk is extracted to determine the attenuation coefficient of the six-degree-of-freedom perturbation vector of the trunk transmitted along each leg joint of the four legs. The attenuation coefficient is labeled as the path feature parameter; Using the six-degree-of-freedom perturbation vector of the torso and the path feature parameters as the basis for map construction, a mapping association table is constructed to record the additional perturbation components experienced by each of the four legs. Then, the reaction force transmission relationship map is generated based on the mapping association table.

4. The method for force-position coordinated control of the quadrupeds and arms of a centaur robot according to claim 1, characterized in that, The process of analyzing the single-leg swing marker or full-support marker carried in the quadrupedal support phase parameters to identify the quadrupedal support state of the centaur robot, and configuring initial stiffness parameters for the ends of the twin arms based on the single-leg swing marker or full-support marker, includes: The single-leg swing marker carried in the quadrupedal support phase parameter is analyzed to detect whether the quadrupedal support state of the centaur robot corresponds to the single-leg swing state. When the single-leg swing marker is detected, the centaur robot is determined to be in a walking state or a posture adjustment state based on the single-leg swing marker. Extract the first impedance reference profile stored in the first data pool, the first impedance reference profile containing the set compliance follow stiffness value; Using the set compliant following stiffness value as the stiffness source corresponding to the single-leg swing mark, the set compliant following stiffness value is assigned to the ends of both arms to generate the initial stiffness parameter corresponding to the single-leg swing mark, thereby enabling the ends of both arms to have compliant characteristics that absorb external vibrations.

5. The method for force-position coordinated control of the quadrupeds and arms of a centaur robot according to claim 4, characterized in that, The step of analyzing the single-leg swing marker or full-support marker carried in the quadrupedal support phase parameters to identify the quadrupedal support state of the centaur robot, and configuring initial stiffness parameters for the ends of the two arms based on the single-leg swing marker or full-support marker, further includes: The full support marker carried in the quadrupedal support phase parameter is analyzed to detect whether the quadrupedal support state of the centaur robot corresponds to the full support state. When the full support marker is detected, it is determined that the Centaur robot is in a stationary working state based on the full support marker; Extract the second impedance reference profile stored in the second data pool. The second impedance reference profile contains positioning maintenance stiffness values ​​configured for positioning operations. The positioning maintenance stiffness values ​​have stronger disturbance rejection properties compared to the compliant following stiffness values. Using the positioning and maintaining stiffness value as the stiffness source corresponding to the full support mark, the positioning and maintaining stiffness value is assigned to the end of the dual arms to generate the initial stiffness parameter corresponding to the full support mark, thereby controlling the positioning accuracy of the end of the dual arms when performing operations.

6. A force-position coordination control device for the quadrupeds and arms of a centaur robot, characterized in that, include: The parameter acquisition module is configured to acquire the quadrupedal support phase parameters and the end-effector operation parameters of the Centaur robot. The closed-loop kinematic chain mapping module is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The relationship graph construction module is configured to construct a reaction force transmission relationship graph that characterizes the force distribution characteristics of the whole machine based on the path feature parameters that characterize the transmission of contact force at the ends of the two arms to the four legs in the closed-loop kinematic chain. The initial stiffness configuration module is configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms according to the single-leg swing mark or full-support mark. The impedance adjustment generation module is configured to compensate the initial stiffness parameter based on the reaction force transmission relationship spectrum to generate a target force potential impedance adjustment parameter for the ends of the double arms. The collaborative instruction generation module is configured to generate posture compensation instructions for driving the joints of both arms according to the target force-position impedance adjustment parameters, and to generate gait stabilization instructions for driving the joints of the four legs after the posture compensation instructions are issued, so that the arms and the four legs perform force-position collaborative operations.

7. An electronic device, characterized in that, Including memory and processor; The memory stores computer programs; When the processor executes the computer program, it implements the method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program; When the computer program is executed by a processor, it implements the method described in any one of claims 1-5.

9. A centaur robot, characterized in that, It includes the robot torso, quadruped mechanism, dual-arm mechanism, sensor acquisition components, dual-arm joint hardware controller, quadruped joint hardware controller, and control processor; The quadruped mechanism is connected to the robot's torso, and the quadruped mechanism includes four leg joints and four leg ends; The dual-arm mechanism is connected to the robot's torso, and the dual-arm mechanism includes dual-arm joints and dual-arm ends; The sensing and acquisition component is configured to acquire the four-leg support phase parameters corresponding to the four-legged mechanism and the end-effector operation parameters corresponding to the two-arm mechanism. The control processor is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The control processor is also configured to construct a reaction force transmission relationship map representing the force distribution characteristics of the whole machine based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs. The control processor is also configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms according to the single-leg swing mark or full-support mark. The control processor is also configured to compensate the initial stiffness parameter based on the reaction force transmission relationship diagram to generate a target force potential impedance adjustment parameter for the ends of the two arms. The control processor is also configured to generate attitude compensation commands for driving the bi-arm joints according to the target force impedance adjustment parameters, and to send the attitude compensation commands to the bi-arm joint hardware controller. The control processor is also configured to generate gait stabilization commands for driving the quadrupedal joints after the attitude compensation command is issued, and to issue the gait stabilization commands to the quadrupedal joint hardware controller so that the dual-arm mechanism and the quadrupedal mechanism perform force-position coordinated operations.

10. A force-position coordination control system for the quadrupedal legs and dual arms of a centaur robot, characterized in that, It includes a centaur robot, a parameter acquisition unit, a collaborative control unit, a dual-arm joint hardware controller, and a quadruped joint hardware controller; The centaur robot includes a quadruped mechanism and a dual-arm mechanism; The parameter acquisition unit is configured to acquire the four-legged support phase parameters corresponding to the four-legged mechanism and the double-arm end operation parameters corresponding to the double-arm mechanism. The collaborative control unit is configured to map the quadrupedal support phase parameters and the end-effector operation parameters of the two arms into a closed-loop kinematic chain. The collaborative control unit is also configured to construct a reaction force transmission relationship map representing the force distribution characteristics of the whole machine based on the path characteristic parameters in the closed-loop kinematic chain that characterize the transmission of contact force at the ends of the two arms to the four legs. The collaborative control unit is also configured to parse the single-leg swing mark or full-support mark carried in the quadrupedal support phase parameter to identify the quadrupedal support state of the Centaur robot, and configure the initial stiffness parameter for the ends of the two arms of the dual-arm mechanism according to the single-leg swing mark or the full-support mark. The collaborative control unit is also configured to compensate the initial stiffness parameter based on the reaction force transmission relationship diagram to generate a target force potential impedance adjustment parameter for the ends of the two arms. The collaborative control unit is also configured to generate attitude compensation commands for driving the dual-arm joints of the dual-arm mechanism according to the target force impedance adjustment parameters, and send the attitude compensation commands to the dual-arm joint hardware controller. The collaborative control unit is also configured to generate gait stabilization commands for driving the quadrupedal joints of the quadrupedal mechanism after the attitude compensation command is issued, and send the gait stabilization commands to the quadrupedal joint hardware controller so that the dual-arm mechanism and the quadrupedal mechanism perform force-position collaborative operations.

Images