Whole-body cooperative jumping control method for quadruped robot with mechanical arms

Abstract

The invention discloses a whole-body cooperative jumping control method for a quadruped robot with mechanical arms, and belongs to the technical field of robot control. The strong coupling effects of a machine body, a mechanical arm, an air phase and an impact phase are integrated into the same model, and the real-time influence of the action of the mechanical arm on the mass center and inertia of the whole machine is quantified; by implementing impact buffering and posture maintaining, the unbalance risk during landing is reduced; the dynamic conflict is reduced by implementing coupling compensation; a whole-body cooperative jump control strategy is precisely guided from the air to landing stable buffering and then to global cooperative compensation, and a complete jump control closed loop is constructed; coupling disturbance is counteracted through feed-forward compensation, real-time errors are fed back and corrected, and the trajectory tracking precision can be effectively improved; through landing impact phase self-adaptive feedback gain adjustment, the attitude oscillation amplitude can be effectively reduced, and the robot is prevented from toppling over; the compound control is combined with the predictability of feedforward and the fault tolerance of feedback.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, specifically to a method for controlling the jumping of a quadruped robot with a robotic arm in a coordinated manner. Background Technology

[0002] A quadruped robot with a robotic arm is a multifunctional intelligent robot system that integrates a robotic arm into a traditional quadruped robot. It combines the excellent terrain adaptability of a quadruped robot with the precise manipulation capabilities of a robotic arm, thereby achieving a composite function of movement and manipulation, and completing more diverse tasks in complex environments.

[0003] While advanced control and planning technologies such as fixed-base robotic arm control, bipedal / wheeled mobile robots, and drones can be referenced when achieving full-body coordinated jumping control in quadruped robots with robotic arms, a series of unique problems and defects still exist in this field. For example, during a jump, the quadruped robot is in the mid-air and impact phase upon landing, and the robotic arm's movements significantly change the position of the robot's center of mass and the distribution of inertia, thus affecting the jump trajectory, attitude stability, and landing safety. Existing control schemes usually decouple the robot body from the robotic arm, such as through layered control, but ignore the strong coupling effect between the two during the jump, leading to landing instability or mission failure. Furthermore, the lack of a unified full-system dynamic model makes it difficult to perform accurate feedforward compensation. Summary of the Invention

[0004] The purpose of this invention is to provide a method for controlling the jumping of a quadruped robot with a robotic arm in a coordinated manner, in order to solve the technical problems mentioned in the background art.

[0005] The objective of this invention can be achieved through the following technical solutions: A method for controlling the jumping of a quadruped robot with a robotic arm in a coordinated manner, comprising: A unified full-system dynamic model of a quadruped robot with a robotic arm is established. Based on the strong coupling effect of the body, robotic arm, aerial phase and landing impact phase, an extended state vector is defined to describe the position of the center of mass, attitude angle, position and velocity of each joint, and lumped disturbance of the whole robot. Based on the established unified whole-system dynamics model, a whole-body coordinated jump control strategy is designed: in the mid-air phase of the jump, the trajectory of the whole machine's center of mass is dynamically adjusted by the joint motion of the robotic arm to correct the deviation of the jump trajectory; in the landing impact phase, the impact load is buffered by the attitude adjustment of the robotic arm to maintain attitude stability, and the coupling effect of the robotic arm's movement on the whole machine's center of mass offset and inertia distribution is calculated and compensated in real time. Based on a unified whole-system dynamics model and a whole-body coordinated jump control strategy, a feedforward-feedback composite control law is designed: the model feedforward is used to compensate for the coupling disturbance between the robotic arm and the body, and the feedback control is combined to dynamically adjust the control input of the quadruped joints and robotic arm joints, and suppress the attitude oscillation of the landing impact phase, so as to ensure the accuracy of jump trajectory tracking and landing safety.

[0006] Furthermore, when integrating the inertial parameters of the fuselage and the robotic arm, the Newton-Euler recursive method is used to derive the dynamic equations of the air phase, involving the following expressions: ;in, To unify the inertia matrix; The Coriolis force and centrifugal force matrix; These are the generalized joint angle vector, the generalized joint angular velocity vector, and the generalized joint angular acceleration vector, respectively. The vector of the gravity term; For generalized control input; This is the lumped disturbance vector; During critical coupling processing, the inertial parameters of link i of the robotic arm are transformed to the body coordinate system using a DH transformation matrix. The inertial parameters include the mass of link i. Location of the center of mass Inertial tensor Real-time updates of the entire machine's center of gravity coordinates .

[0007] Furthermore, for the instantaneous contact between the legs, robotic arm, and the ground, the Hertzian contact force model is introduced to describe the contact force. The dynamic equations of the impact phase are derived using the momentum theorem, and the relevant expressions are: ; ;in, This is the contact force vector; This is the Hertzian stiffness coefficient; This refers to the amount of contact deformation. The exponent for the Hertzian elasticity term; This is the contact damping coefficient; The contact deformation rate; This refers to the change in joint angular velocity at the moment of impact. The Jacobian matrix for the contact area; Add a disturbance vector to the impact.

[0008] Furthermore, the acquired air phase data and impact phase state data are integrated to define an extended state vector X: ;in, These are the joint position vector, velocity vector, and acceleration vector, respectively. The coordinates of the machine's center of mass; This is the lumped disturbance vector.

[0009] Furthermore, when dynamically adjusting the trajectory of the entire machine's center of mass through the joint motion of the robotic arm to correct the jump trajectory deviation, the current state vector X and the target center of mass trajectory are input. And calculate the centroid deviation. The expression involved is: ; Let be the actual centroid position of the robot at time t; And, utilizing the Jacobian matrix of the robotic arm Calculate the joint adjustment of the robotic arm The expression involved is: ;in, For the transpose of the Jacobian matrix of the robotic arm, This is the proportional gain matrix; Send the robotic arm joint adjustment command and output the robotic arm joint adjustment amount. .

[0010] Furthermore, when implementing pre-landing contact prediction, the current airspeed is obtained. and current height And obtain the predicted landing time through calculation. Predicting landing contact points The expression involved is: ;in, The velocity is in the vertical direction; ;in, Let t be the position of the robot's contact point in the x-direction at the current time; Let t be the velocity of the robot's contact point in the x-direction at the current time t; This is the predicted time interval from the current time t to the moment of landing; Let t be the position of the robot's contact point in the y-direction at the current time. Let t be the velocity of the robot's contact point in the y-direction at the current time. This represents the z-coordinate of the point of contact with the ground.

[0011] Furthermore, during the attitude adjustment of the impact-phase robotic arm, the contact point is obtained. Impact angular velocity And calculate the contact deformation. ;in, This indicates the incident velocity of the contacting body in the direction perpendicular to the ground before contact occurs; For equivalent contact quality; Based on the dynamic equation of the impact phase Adjust the extension of the robotic arm end forward by M meters, where M is a preset value, and simultaneously adjust the preload of the leg joints to provide cushioning.

[0012] Furthermore, during real-time coupling compensation, the joint angles of the robotic arm after adjustment are obtained. The inertial parameters of link i of the robotic arm are transformed to the current body coordinate system using the DH transformation; and the center of mass of the whole machine is calculated and updated, and the joint torque is compensated.

[0013] Furthermore, when using model feedforward compensation to offset the coupling disturbance between the robotic arm and the fuselage, the feedforward compensation term is pre-calculated to counteract the coupling disturbance between the robotic arm and the fuselage, reducing the burden on feedback control. The relevant expression is as follows: ;in, This is the feedforward control input vector; The desired joint acceleration vector; Coriolis / centrifugal force matrix; The desired joint angular velocity vector; This is the vector of the coupling perturbation term.

[0014] Furthermore, a feedback control term is designed to address the impact phase error during landing, using the robot's sensors to collect real-time data on the actual joint angles. Actual joint angular velocity Actual fuselage attitude angle Actual fuselage attitude angular velocity The sensors include, but are not limited to, joint encoders and IMU sensors. The feedback control input vector is calculated and used to correct trajectory tracking errors. The relevant expressions are as follows: ;in, For feedback control input vector; This is the joint position feedback gain matrix; This is the joint angle error vector. , These are the desired joint angle and the actual joint angle, respectively. This is the joint angular velocity feedback gain matrix; This is the joint angular velocity error vector. , These are the desired joint angular velocity and the actual joint angular velocity, respectively. The fuselage attitude angle feedback gain matrix; This is the fuselage attitude angle error vector. , These are the desired fuselage attitude angle and the actual fuselage attitude angle, respectively. The gain matrix for fuselage attitude angular velocity feedback; Let be the fuselage attitude angular velocity error vector. , These are the desired fuselage attitude angular velocity and the actual fuselage attitude angular velocity, respectively. The feedforward compensation term and the feedback control term are linearly superimposed to obtain the final control input of the quadrupedal joint and the robotic arm joint, and the final control input is sent to the joint actuator.

[0015] Compared to existing solutions, the beneficial effects achieved by this invention are: This invention integrates the strong coupling effects of the fuselage, robotic arm, air phase, and impact phase into a single model, enabling the real-time quantification of the robotic arm's movements on the overall center of mass and inertia, effectively solving the problem of neglected coupling. By extending the state vector, it can provide complete state inputs for subsequent coordinated control, thus supporting precise control.

[0016] This invention effectively reduces landing position deviation by implementing aerial trajectory correction, meeting the requirements of high-precision jumping tasks; it reduces the risk of imbalance upon landing and enhances landing stability by implementing impact buffering and attitude maintenance; it reduces dynamic conflicts by implementing coupling compensation, enabling control commands to be more efficiently converted into precise actions; the whole-body coordinated jumping control strategy constructs a complete jumping control closed loop from precise aerial guidance to smooth landing buffering and then to global coordinated compensation, providing core support for robots to efficiently complete jumping tasks.

[0017] This invention effectively improves trajectory tracking accuracy by using feedforward compensation to offset coupling disturbances and feedback to correct real-time errors. Adaptive feedback gain adjustment during the impact phase effectively reduces attitude oscillation amplitude, preventing robot tipping. The composite control, combining the predictive nature of feedforward with the fault tolerance of feedback, adapts to impact variations in ground hardness, effectively enhancing control robustness. Strict termination conditions and reaction force monitoring ensure the robot structure remains undamaged. Attached Figure Description

[0018] The invention will now be further described with reference to the accompanying drawings.

[0019] Figure 1 This is a flowchart of a method for controlling the jumping of a quadruped robot with a robotic arm in a coordinated manner according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, this invention is a method for coordinated jumping control of a quadruped robot with a robotic arm, comprising: A unified full-system dynamic model of a quadruped robot with a robotic arm is established. Based on the strong coupling effect of the body, robotic arm, aerial phase, and landing impact phase, an extended state vector is defined to describe the position of the robot's center of mass, attitude angles, joint positions and velocities, and lumped disturbances. Specific steps include: When constructing a unified full-system dynamic model, an inertial coordinate system O-XYZ, a fuselage body coordinate system B-xyz, and a robotic arm link coordinate system Li-xyz are constructed based on a quadrupedal fuselage and a 6-DOF robotic arm. The quadrupedal body consists of four legs, each with three joints; the six-DOF robotic arm consists of seven links and six joints. The inertial coordinate system is fixed to the ground, with the origin being the ground reference point; The origin of the fuselage body coordinate system is located at the fuselage center of mass, and the coordinate axes are aligned with the fuselage structure; The robotic arm link coordinate system describes the pose of the i-th link relative to the previous link using DH parameters; the DH parameters include the link length. Twist angle offset and joint angle ; When integrating the inertial parameters of the fuselage and the robotic arm, the Newton-Euler recursive method is used to derive the dynamic equations of the airborne phase. Since the airborne phase has no ground contact, the relevant expressions are: ;in, The generalized joint angle vector is specifically 4 legs × 3 joints plus 6 joints for the robotic arm. ; To unify the inertia matrix, The fuselage inertial matrix Leg inertia matrix Robotic arm link inertia matrix It is formed by stacking. The joint angles of the robotic arm affect the inertial distribution in real time; This is the Coriolis force and centrifugal force matrix, which includes the Coriolis force and centrifugal force generated during system motion. The Coriolis force is cross-correlated with velocity and acceleration, while the centrifugal force is related to the square of the velocity. ; These are the generalized joint angular velocity vector and the generalized joint angular acceleration vector, respectively, both with dimension 1. ; The vector of the gravity term. ; The generalized control input includes the leg joint torque plus the robotic arm joint driving force. ; This is the lumped perturbation vector. This includes all undesirable external disturbances, such as inertial changes caused by robotic arm movements, air resistance, model errors, etc., which need to be eliminated through control compensation. It should be noted that by calculating the forward recursion of the speed, acceleration, and inertial force of each link in the computer body, robotic arm, and legs, and the reverse recursion of the joint torque, and by integrating the whole body inertial parameters, a unified aerial phase dynamics model can be constructed, which can completely describe the motion law of the robot when there is no ground contact. Its functions include: eliminating the dynamic decoupling between the body and the robotic arm, achieving an accurate description of the whole-body coordinated motion, providing a core basis for aerial phase trajectory optimization, such as center of mass trajectory planning and robotic arm posture adjustment; in addition, the recursive dynamics calculation efficiency of the Newton-Euler method is high, adapting to the dynamic response requirements of robots and supporting real-time control; and it is compatible with multi-link structures, including the body, 6-DOF robotic arm, and quadruped legs, and can be extended to complex robot systems.

[0022] During critical coupling processing, the inertial parameters of link i of the robotic arm are transformed to the body coordinate system using a DH transformation matrix. The inertial parameters include the mass of link i. Location of the center of mass Inertial tensor Real-time updates of the entire machine's center of gravity coordinates The expression involved is: ;in, For fuselage weight; 1 represents the center of mass of the robot body; i represents the link number, i from 1 to 7 corresponds to the 7 links of the 6-DOF robot arm, that is, 6 joints connect 7 links, covering all moving parts of the robot arm; Let be the position vector of the centroid of the i-th link of the robotic arm in the body coordinate system. The DH transformation matrix is ​​used to transform the fixed position of the centroid of link i in its local coordinate system to the body coordinate system. Because the joint angle q of the robotic arm changes, the DH transformation matrix is ​​updated in real time. It changes dynamically with q; Let be the position vector of the fuselage's center of mass in the fuselage coordinate system; Indicates the total mass of the machine; It should be noted that the inertial parameters of each link of the robotic arm are transformed to the body coordinate system through the DH transformation matrix. After unifying the reference system, the coordinates of the whole machine's center of mass are calculated and dynamically updated as the joint angle changes. Its functions include: breaking down the coordinate system barriers between the robotic arm and the robot body, achieving global unification of inertial parameters, and avoiding deviations in the calculation of the center of mass due to the movement of the robotic arm; real-time updated coordinates of the entire robot's center of mass provide key feedback for aerial phase balance control, such as adjusting the leg posture to maintain the stability of the center of mass, which can prevent the robot from becoming unbalanced; and the standardized coordinate transformation rules of the DH transformation can ensure the accuracy of inertial parameter mapping, laying a reliable foundation for subsequent dynamic calculations.

[0023] For the instantaneous contact between the legs / robotic arm and the ground, the Hertzian contact force model is introduced to describe the contact force. The dynamic equations of the impact phase are derived using the momentum theorem. The relevant expressions are: ; ;in, The contact force vector has a dimension of The force exerted by the ground on the robot's contact points, such as the reaction force when the end of the legs and the robotic arm links come into contact with the ground; This is the Hertzian stiffness coefficient. R is the equivalent radius of the contact body. , These are the radii of curvature of the robot's contact area and the radius of curvature of the ground, respectively. For the equivalent elastic modulus, , These are the elastic modulus of the robot's contact area and the elastic modulus of the ground, respectively. These are the Poisson's ratios of the robot's contact points and the ground, respectively. This refers to the amount of contact deformation, specifically the relative compression between the contact area of ​​a leg or foot, or the end of a robotic arm, and the ground. A positive number indicates that contact deformation has occurred. The exponent for the Hertzian elasticity term; The contact damping coefficient describes the energy dissipation during the contact process, such as vibration attenuation during impact. The calculation formula is: , The contact equivalent mass is specifically the weighted mass of the robot's contact area relative to the ground. The damping ratio, ranging from 0.1 to 0.3, is determined by the damping characteristics of the material. The contact deformation velocity is specifically the time derivative of the contact deformation, which reflects the speed at which the contact part is pressed into the ground. It is positive at the moment of impact. The change in joint angular velocity at the moment of impact represents the difference in joint angular velocity before and after the impact. The Jacobian matrix for the contact area has dimensions of . Map the joint space velocity to the Cartesian space velocity at the contact point; An additional perturbation vector is added to the impact, with dimension . This indicates additional disturbances introduced during the impact process, such as inelastic collision losses at the contact points or inertial impacts caused by sudden changes in the robotic arm's posture. It should be noted that the Hertz model is used to quantify the contact force, that is, the nonlinear elastic-damped force reflects the soft and hard characteristics of the contact. Combined with the momentum theorem, that is, the change in momentum at the moment of impact is equal to the impulse, the dynamic equation of the impact phase is derived, which can describe the relationship between the force and motion changes at the moment of contact. Its functions include: Hertzian model captures the nonlinear characteristics of contact force, reflecting the ground contact process more realistically than linear spring model; momentum theorem ignores the instantaneous displacement change during impact and focuses on the sudden change in velocity, which can accurately describe the impact characteristics of short time and large load; and it can provide quantitative basis for impact buffering strategies such as energy absorption by robotic arm posture adjustment and torque compensation of leg joints, protecting the robot structure and maintaining stability after impact.

[0024] The above steps, from aerial phase modeling and whole-body coupling integration to impact phase quantization, form a complete closed loop for robot dynamic control, supporting precise control of whole-body coordinated motion.

[0025] The acquired air phase data and impact phase state data are integrated to define an extended state vector X: ;in, These are the joint position vector, velocity vector, and acceleration vector, respectively. The coordinates of the machine's center of mass; This is the lumped disturbance vector.

[0026] In this embodiment of the invention, by integrating the strong coupling effects of the fuselage, robotic arm, air phase, and impact phase into the same model, the real-time impact of the robotic arm's movements on the overall center of mass and inertia can be quantified, effectively solving the problem of neglecting coupling. By extending the state vector, complete state inputs can be provided for subsequent cooperative control, enabling precise control. Furthermore, the above steps can also provide a unified dynamic basis for subsequent cooperative strategies and feedforward compensation.

[0027] Based on the established unified whole-system dynamics model, a whole-body coordinated jump control strategy is designed: In the air phase of the jump, the trajectory of the entire machine's center of gravity is dynamically adjusted through the joint motion of the robotic arm to correct the jump trajectory deviation; in the landing impact phase, the impact load is buffered through the attitude adjustment of the robotic arm to maintain attitude stability, and the coupled effects of the robotic arm's movements on the machine's center of gravity offset and inertia distribution are calculated and compensated in real time; the specific steps include: Get the robot's current joint angle Joint angular velocity Initial position of the machine's center of gravity Target jump trajectory parameters; among which, the target jump trajectory parameters include height Landing location , , The distance on the X-axis relative to the origin O is the projection of the robot's center of gravity onto the ground at the target landing position. The distance of the center of gravity projection point of the robot's target landing position relative to the origin O in the Y-axis direction; these two parameters together define the horizontal target landing point of the robot's jumping task and are key inputs for subsequent trajectory planning and control strategies; , , The default values ​​can be 0.3, 1, or 0, all in meters; Load the inertial parameters of the unified dynamics model and generate the initial state vector. ;in, Initially, all values ​​are 0; When dynamically adjusting the trajectory of the robot's center of mass by using the joint motion of the robotic arm to correct jump trajectory deviations, the current state vector X and the target center of mass trajectory are acquired and input. t is a time variable; Based on the current state X and the target centroid trajectory Calculate the centroid deviation The expression involved is: ; Let be the actual centroid position of the robot at time t; And, utilizing the Jacobian matrix of the robotic arm Calculate the joint adjustment of the robotic arm Jacobian matrix of robotic arm For a 6×6 dimension, mapping the joint space to Cartesian space involves the following expressions: ;in, For the transpose of the Jacobian matrix of the robotic arm, The proportional gain matrix can be specifically defined as follows: The first three diagonal elements correspond to the proportional gains of the x, y, and z position degrees of freedom. The larger the value, the more sensitive the correction of position deviation. The last three diagonal elements correspond to the proportional gains of the roll, pitch, and yaw attitude degrees of freedom. The smaller the value, the lower the strength of attitude correction is, which avoids instability caused by excessive attitude adjustment. Send the robotic arm joint adjustment command and output the robotic arm joint adjustment amount. ; It should be noted that by using the joint motion of the robotic arm to dynamically adjust the trajectory of the entire machine's center of gravity and actively correct trajectory deviations during the jump, it is possible to ensure that the landing position meets expectations.

[0028] When conducting pre-landing contact prediction, obtain the current air speed. and current height And obtain the predicted landing time through calculation. Predicting landing contact points The expression involved is: ;in, The vertical velocity is obtained via an IMU sensor. ;in, Let t be the position of the robot's contact point in the x-direction at the current time; Let t be the velocity of the robot's contact point in the x-direction at the current time t; This is the predicted time interval from the current time t to the moment of landing; Let t be the position of the robot's contact point in the y-direction at the current time. Let t be the velocity of the robot's contact point in the y-direction at the current time. This represents the z-coordinate of the point of contact with the ground. When performing attitude adjustment of the robotic arm during impact phase, the contact point is obtained. Impact angular velocity And calculate the contact deformation. ;in, This indicates the incident velocity of the contacting body in the direction perpendicular to the ground before contact occurs; the contacting body is, for example, the end of a robot's foot. Equivalent contact mass means that the masses of multiple objects involved in the contact (such as the robot body and foot structure) are equivalent to a single mass. This is used to simplify the contact dynamics model and ignore the mass influence of non-key components. Based on the dynamic equation of the impact phase Adjust the extension of the robotic arm end to extend forward by M meters, where M is a preset value with a default value of 0.2, and simultaneously adjust the preload of the leg joints to provide cushioning. It should be noted that by adjusting the posture of the robotic arm, the impact load can be dispersed, reducing the damage to the robot body caused by the ground reaction force. At the same time, it can maintain the stability of the overall posture and prevent tipping.

[0029] When implementing real-time coupling compensation, the joint angles of the robotic arm after adjustment are obtained. In the current fuselage coordinate system, the inertial parameters of link i of the robotic arm are transformed to the fuselage coordinate system through DH transformation; In addition, the calculation and update of the overall centroid involves the following expression: ;in, This is the updated coordinate vector of the machine's centroid. For fuselage weight; Let be the coordinate vector of the fuselage's center of mass in the fuselage coordinate system; Let i be the mass of the robotic arm link i; Let be the coordinate vector of the centroid of the i-th link of the robotic arm in the body coordinate system; In addition, the expression for compensating for joint torques is as follows: ;in, The compensated joint torque; The original joint torque; For coupling interference torque; When judging the deviation of the center of gravity after landing, if and If so, then control will be terminated; This represents the difference between the actual coordinates of the robot's center of mass in the x-direction and the target coordinates after landing. C0 is the difference between the actual coordinates of the robot's center of mass in the y-direction and the target coordinates after landing; C0 is the preset allowable threshold for center of mass deviation, with a default value of 0.05m. Otherwise, repeat the above steps to continue adjusting.

[0030] It should be noted that by calculating the coupling effect of the robotic arm's movements with the overall machine's center of gravity shift and inertia distribution in real time and performing dynamic compensation, the interference of local movements on the overall dynamic balance can be eliminated, ensuring the coordinated and consistent operation of the entire machine.

[0031] In this embodiment of the invention, by implementing aerial trajectory correction, the landing position deviation can be effectively reduced to meet the requirements of high-precision jumping tasks; by implementing impact buffering and attitude maintenance, the risk of imbalance during landing is reduced and landing stability is enhanced; by implementing coupling compensation, dynamic conflicts are reduced, enabling control commands to be more efficiently converted into precise actions; the whole-body coordinated jumping control strategy constructs a complete jumping control closed loop from precise aerial guidance to smooth landing buffering and then to global coordinated compensation, providing core support for the robot to efficiently complete jumping tasks.

[0032] Based on a unified whole-system dynamics model and a whole-body coordinated jump control strategy, a feedforward-feedback composite control law is designed: model feedforward is used to compensate for coupling disturbances between the robotic arm and the body, combined with feedback control to dynamically adjust the control inputs of the quadrupedal joints and robotic arm joints, and to suppress attitude oscillations during the landing impact phase, ensuring jump trajectory tracking accuracy and landing safety. Specific steps include: When using model feedforward compensation to offset the coupling disturbance between the robotic arm and the fuselage, the feedforward compensation term is pre-calculated to counteract the coupling disturbance and reduce the burden on feedback control. The relevant expression is as follows: ;in, This is the feedforward control input vector; The desired joint acceleration vector; Coriolis / centrifugal force matrix; The desired joint angular velocity vector; This is the vector of coupling perturbation terms; Specifically, the feedforward control input vector for each control cycle is obtained through offline pre-calculation or real-time recursion. , which serves as the fundamental component of the control input.

[0033] To address the impact phase error during landing, a feedback control term is designed, which uses the robot's sensors to collect the actual joint angles in real time. Actual joint angular velocity Actual fuselage attitude angle Actual fuselage attitude angular velocity The sensors include, but are not limited to, joint encoders and IMU sensors. The feedback control input vector is calculated and used to correct trajectory tracking errors. The relevant expressions are as follows: ;in, For feedback control input vector; This is the joint position feedback gain matrix; This is the joint angle error vector. , These are the desired joint angle and the actual joint angle, respectively. This is the joint angular velocity feedback gain matrix; This is the joint angular velocity error vector. , These are the desired joint angular velocity and the actual joint angular velocity, respectively. The fuselage attitude angle feedback gain matrix; This is the fuselage attitude angle error vector. , These are the desired fuselage attitude angle and the actual fuselage attitude angle, respectively. The gain matrix for fuselage attitude angular velocity feedback; Let be the fuselage attitude angular velocity error vector. , These are the desired fuselage attitude angular velocity and the actual fuselage attitude angular velocity, respectively. The feedforward compensation term and the feedback control term are linearly superimposed to obtain the final control input for the quadrupedal joint and the robotic arm joint. This final control input is then sent to the joint actuator. The relevant expressions are: ;in, For final control input; When suppressing attitude oscillations during the landing impact phase, if the force sensor detects that the ground reaction force exceeds a preset threshold (e.g., 1.5 times the robot's own weight), adaptive parameter adjustment is initiated. The relevant expression is: ; ;in, The adaptively adjusted fuselage attitude angle position feedback gain matrix; This is the adaptive coefficient for attitude angle error, with a value ranging from 0.1 to 0.3. This is the vector of absolute values ​​of the attitude angle error; The adaptively adjusted fuselage attitude angular velocity feedback gain matrix; This is the adaptive coefficient for attitude angular velocity error, with a value ranging from 0.2 to 0.4. This is the vector of absolute values ​​of the attitude angular velocity error; The adaptively adjusted fuselage attitude angular position feedback gain matrix and the adaptively adjusted fuselage attitude angular velocity feedback gain matrix are used to suppress attitude oscillations caused by shocks. When confirming whether the control effect has met the standards, monitor and analyze the judgment criteria: Jump trajectory tracking error standard: and ; Attitude oscillation amplitude standard: and ; Landing safety standard: The peak ground reaction force is less than or equal to twice the robot's own weight; If all criteria are met, terminate composite control; Otherwise, recalculate the feedforward term and continue adjusting the control input.

[0034] It should be noted that by using feedforward compensation to offset coupling disturbances and feedback to correct real-time errors, the trajectory tracking accuracy can be effectively improved; by using adaptive feedback gain adjustment for the landing impact phase, the attitude oscillation amplitude can be effectively reduced, preventing the robot from tipping over; the composite control combines the predictive nature of feedforward with the fault tolerance of feedback to adapt to impact changes with different ground hardness, effectively enhancing the robustness of the control; and by implementing strict termination conditions and reaction force monitoring, the robot structure is protected from damage.

[0035] The above steps, through a logic chain of feedforward compensation for coupled disturbances, feedback error correction, composite fusion control, impact adaptive adjustment, and closed-loop verification, achieve precise and stable control of jumping tasks, providing core technical support for complex motion tasks of quadruped robots equipped with robotic arms.

[0036] In the several embodiments provided by this invention, it should be understood that the disclosed system can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for example, the division of modules is only a logical functional division, and there may be other division methods in actual implementation.

[0037] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0038] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in the form of hardware plus software functional modules.

[0039] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the essential characteristics of the present invention.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for controlling the jumping of a quadruped robot with a robotic arm in a coordinated manner, characterized in that, include: A unified full-system dynamic model of a quadruped robot with a robotic arm is established. Based on the strong coupling effect of the body, robotic arm, aerial phase and landing impact phase, an extended state vector is defined to describe the position of the center of mass, attitude angle, position and velocity of each joint, and lumped disturbance of the whole robot. Based on the established unified whole-system dynamics model, a whole-body coordinated jump control strategy is designed: in the mid-air phase of the jump, the trajectory of the whole machine's center of mass is dynamically adjusted by the joint motion of the robotic arm to correct the deviation of the jump trajectory; in the landing impact phase, the impact load is buffered by the attitude adjustment of the robotic arm to maintain attitude stability, and the coupling effect of the robotic arm's movement on the whole machine's center of mass offset and inertia distribution is calculated and compensated in real time. Based on a unified whole-system dynamics model and a whole-body coordinated jump control strategy, a feedforward-feedback composite control law is designed: the model feedforward is used to compensate for the coupling disturbance between the robotic arm and the body, and the feedback control is combined to dynamically adjust the control input of the quadruped joints and robotic arm joints, and suppress the attitude oscillation of the landing impact phase, so as to ensure the accuracy of jump trajectory tracking and landing safety.

2. The method for whole-body coordinated jumping control of a quadruped robot with a robotic arm according to claim 1, characterized in that, When integrating the inertial parameters of the fuselage and the robotic arm, the Newton-Euler recursive method is used to derive the dynamic equations of the air phase, involving the following expressions: ;in, To unify the inertia matrix; The Coriolis force and centrifugal force matrix; These are the generalized joint angle vector, the generalized joint angular velocity vector, and the generalized joint angular acceleration vector, respectively. The vector of the gravity term; For generalized control input; This is the lumped disturbance vector; During critical coupling processing, the inertial parameters of link i of the robotic arm are transformed to the body coordinate system using a DH transformation matrix. The inertial parameters include the mass of link i. Location of the center of mass Inertial tensor Real-time updates of the entire machine's center of gravity coordinates .

3. The method for controlling the jumping of a quadruped robot with a robotic arm in full-body coordinated movement according to claim 2, characterized in that, For the instantaneous contact between the legs / robotic arm and the ground, the Hertzian contact force model is introduced to describe the contact force. The dynamic equations of the impact phase are derived using the momentum theorem. The relevant expressions are: ； ;in, This is the contact force vector; This is the Hertzian stiffness coefficient; This refers to the amount of contact deformation. The exponent for the Hertzian elasticity term; This is the contact damping coefficient; The contact deformation rate; This refers to the change in joint angular velocity at the moment of impact. The Jacobian matrix for the contact area; Add a disturbance vector to the impact.

4. The method for controlling the jumping of a quadruped robot with a robotic arm in full-body coordinated movement according to claim 3, characterized in that, The acquired air phase data and impact phase state data are integrated to define an extended state vector X: ;in, These are the joint position vector, velocity vector, and acceleration vector, respectively. The coordinates of the machine's center of mass; This is the lumped disturbance vector.

5. A method for controlling the jumping of a quadruped robot with a robotic arm in full-body coordinated movement according to claim 4, characterized in that, When dynamically adjusting the overall center of mass trajectory of the robotic arm through joint motion to correct jump trajectory deviation, the inputs are the current state vector X and the target center of mass trajectory. And calculate the centroid deviation. The expression involved is: ; Let be the actual centroid position of the robot at time t; And, utilizing the Jacobian matrix of the robotic arm Calculate the joint adjustment of the robotic arm The expression involved is: ;in, This is the transpose of the Jacobian matrix of the robotic arm; This is the proportional gain matrix; Send the robotic arm joint adjustment command and output the robotic arm joint adjustment amount. .

6. A method for controlling the jumping of a quadruped robot with a robotic arm in full-body coordinated movement according to claim 5, characterized in that, When conducting pre-landing contact prediction, obtain the current air speed. and current height And obtain the predicted landing time through calculation. Predicting landing contact points The expression involved is: ;in, The velocity is in the vertical direction; ;in, Let t be the position of the robot's contact point in the x-direction at the current time; Let t be the velocity of the robot's contact point in the x-direction at the current time t; This is the predicted time interval from the current time t to the moment of landing; Let t be the position of the robot's contact point in the y-direction at the current time. Let t be the velocity of the robot's contact point in the y-direction at the current time. This represents the z-coordinate of the point of contact with the ground.

7. A method for controlling the jumping of a quadruped robot with a robotic arm in full-body coordinated movement according to claim 6, characterized in that, When performing attitude adjustment of the robotic arm during impact phase, the contact point is obtained. Impact angular velocity And calculate the contact deformation. ;in, This indicates the incident velocity of the contacting body in the direction perpendicular to the ground before contact occurs; For equivalent contact quality; Based on the dynamic equation of the impact phase Adjust the extension of the robotic arm end forward by M meters, where M is a preset value, and simultaneously adjust the preload of the leg joints to provide cushioning.

8. A method for controlling the jumping of a quadruped robot with a robotic arm in full-body coordinated movement according to claim 7, characterized in that, When implementing real-time coupling compensation, the joint angles of the robotic arm after adjustment are obtained. In the current fuselage coordinate system, the inertial parameters of link i of the robotic arm are transformed to the fuselage coordinate system through DH transformation; In addition, the machine's center of gravity is calculated and updated, and joint torques are compensated.

9. A method for controlling the jumping of a quadruped robot with a robotic arm in full-body coordinated movement according to claim 8, characterized in that, When using model feedforward compensation to offset the coupling disturbance between the robotic arm and the fuselage, the feedforward compensation term is pre-calculated to counteract the coupling disturbance and reduce the burden on feedback control. The relevant expression is as follows: ;in, This is the feedforward control input vector; The desired joint acceleration vector; Coriolis / centrifugal force matrix; The desired joint angular velocity vector; This is the vector of the coupling perturbation term.

10. A method for controlling the jumping of a quadruped robot with a robotic arm in full-body coordinated movement according to claim 9, characterized in that, To address the impact phase error during landing, a feedback control term is designed, which uses the robot's sensors to collect the actual joint angles in real time. Actual joint angular velocity Actual fuselage attitude angle Actual fuselage attitude angular velocity The sensors include, but are not limited to, joint encoders and IMU sensors. The feedback control input vector is calculated and used to correct trajectory tracking errors. The relevant expressions are as follows: ;in, For feedback control input vector; This is the joint position feedback gain matrix; This is the joint angle error vector. , These are the desired joint angle and the actual joint angle, respectively. This is the joint angular velocity feedback gain matrix; This is the joint angular velocity error vector. , These are the desired joint angular velocity and the actual joint angular velocity, respectively. This is the fuselage attitude angle feedback gain matrix; Let be the fuselage attitude angle error vector. , These are the desired fuselage attitude angle and the actual fuselage attitude angle, respectively. The gain matrix for fuselage attitude angular velocity feedback; Let be the fuselage attitude angular velocity error vector. , These are the desired fuselage attitude angular velocity and the actual fuselage attitude angular velocity, respectively. The feedforward compensation term and the feedback control term are linearly superimposed to obtain the final control input of the quadrupedal joint and the robotic arm joint, and the final control input is sent to the joint actuator.

Images