Flying humanoid robot and joint deformation control method and thrust control method thereof

Abstract

This invention relates to the field of electronic sensing information extraction and processing technology, and discloses a flying humanoid robot and its joint deformation control method and thrust control method. The flying humanoid robot of this invention includes a waist base, an upper body frame, a power module and a control device. The control device stores the joint deformation control method and the thrust control method. The method can change the joint rotation angle according to the current state of the robot, thereby changing the thruster layout so that the thrust direction is as far as possible towards the direction of movement, thereby further improving the flight performance without changing the number and power of the robot's existing thrusters.

Description

Technical Field

[0001] This invention relates to the field of electronic sensing information extraction and processing technology, and more specifically, to a flying humanoid robot and its joint deformation control method and thrust control method. Background Technology

[0002] In recent years, flying robots have been widely used in exploration, rescue, and other fields due to their excellent maneuverability and flexibility. Robots can quickly enter work sites, assisting or even replacing humans in carrying out tasks. As research deepens, in addition to common quadcopter drones, researchers are attempting to combine propellers and rotors with other wheeled and legged robots, enabling these robots to gain flight capabilities in addition to low-power ground mobility to overcome more complex terrain limitations. Current research on multi-joint flying robot control methods focuses heavily on controlling propeller output, with little attention paid to joint movement during flight. Therefore, the robot's propeller layout is often fixed, and the only way to improve the robot's mobility is by increasing propeller power. However, improving flight performance by controlling propeller output is limited by the robot's own performance, with an upper limit to performance enhancement.

[0003] Existing technology discloses a control system for controlling a flying robot, relating to the field of flight control technology. This invention addresses the technical problems of flying robots lacking multi-dimensional data for fault prediction of different categories, and failing to provide reasonable flight control operation suggestions for priority or non-priority faults, resulting in insufficient accuracy in fault diagnosis and performance optimization analysis capabilities. By collecting and analyzing data from multiple dimensions, including flight performance, flight control, and external interference, the system can not only feed back multi-category, multi-level signals to the user's remote control module, facilitating user optimization of the robot's flight performance and control accuracy, but also provide strategies for accurate fault diagnosis, structural health monitoring, and performance optimization analysis and control by combining multi-dimensional data for fault prediction of different categories. While this invention overcomes the limitations of relying solely on thrusters to improve flight performance by combining multi-dimensional data for fault prediction of different categories, the improvement in flight performance remains limited and cannot significantly enhance overall flight performance. Summary of the Invention

[0004] The purpose of this invention is to disclose a flying humanoid robot with better flight performance, as well as its joint deformation control method and thrust control method.

[0005] To achieve the above objectives, the present invention provides a flying humanoid robot. The specific technical solution includes a waist base, an upper body frame, a power module, and a control device. The upper body frame is connected to the waist base, and the power module and control device are mounted on the upper body frame. The upper body frame has symmetrical, identical two-degree-of-freedom shoulder modules on both sides. Two leg structures are symmetrically connected to both sides of the waist base. Each leg structure consists of four modules connected in sequence: a hip joint module, a thigh module, a lower leg module, and a foot module. The thigh module and the lower leg module are hinged via a knee joint module, and the lower leg module and the foot module are hinged via an ankle joint module. The first propulsion module is fixed to the back of the dual-degree-of-freedom shoulder module, and the second propulsion module is fixed to the side of the foot module, moving together with the foot module. A drive motor is installed on the waist base, and the output end of the drive motor is connected to the upper frame. The drive motor drives the relative rotation between the upper frame and the waist base. The power module supplies power to the control device and the drive devices of the dual-degree-of-freedom shoulder module, hip joint module, thigh module, calf module, and foot module respectively. The control device is electrically connected to the dual-degree-of-freedom shoulder module, hip joint module, thigh module, calf module, foot module, knee joint module, and ankle joint module respectively.

[0006] Furthermore, this invention also provides a method for controlling joint deformation in a flying humanoid robot, comprising the following steps:

[0007] S1: Obtain the rotation angle information of the shoulder module, hip joint module, knee joint module, and ankle joint module of the flying humanoid robot at the current moment; obtain the thrust magnitude of the shoulder module and foot module and the expected and actual state values ​​of the center of mass of the flying humanoid robot;

[0008] S2: Obtain the weighting coefficients based on the expected state value and the actual state value of the current centroid;

[0009] S3: Based on the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, set the constraint inequality Q for the output. u Q l Furthermore, based on the thrust magnitude of the two-degree-of-freedom shoulder module and foot module at the current moment, as well as the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, and the weight coefficients, a loss function L2 is constructed.

[0010] S4: Based on the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, and constraint inequality Q u Q l And the L2 loss function is used to obtain the rotation information of the target hip joint module, knee joint module, and ankle joint module;

[0011] S5: Perform linear discretization on the target hip joint module, knee joint module, and ankle joint module, and use the corresponding discrete value selected at the current moment as the rotation angle information of the hip joint module, knee joint module, and ankle joint module in the new state;

[0012] S6: Transmit the rotation angle information of the new state hip joint module, knee joint module, and ankle joint module to the control device to control the hip joint module, knee joint module, and ankle joint module.

[0013] Further, in step S2, based on the current desired state value and actual state value of the robot's center of mass, the current center of mass pose error, current center of mass velocity error, and current center of mass angular velocity error are calculated. These errors are then input into the PD controller in the flying humanoid robot control device to calculate weighting coefficients. The specific formula is shown below:

[0014] w=P((p,Θ)-(p des Θ des ))+D((v,ω)-(v des ω des ))

[0015] Where w is the weight vector, P and D are diagonal matrices, with the diagonal elements representing the PD coefficients of the corresponding terms, p is the position of the robot's center of mass in the world frame, Θ is the Euler angle of the robot's center of mass in the world frame, v is the velocity of the robot's center of mass in the world frame, and ω is the angular velocity of the robot's center of mass. des Let ω be the expected velocity of the robot's center of mass in the world frame. des Let be the expected value of the body angular velocity of the robot's center of mass.

[0016] Furthermore, the specific formula for the constraint inequality of the output quantity in step S3 is as follows:

[0017] Qu = min(q) max ,qi-1+ωq max ·△T), Ql= max (q min ,qi1-ωq max ·△T), q max q min These represent the maximum and minimum rotation angles of the hip, knee, and ankle joint modules, respectively, determined by the robot's structure. ωq max ΔT represents the upper limit speed of the joint, and ΔT represents the period of the controller output.

[0018] Furthermore, the specific formula for constructing the loss function L2 in step S3 is as follows:

[0019] L2 = -wT v(qi)+a1||qi-1-qi||2+a2||qi-qd| |2

[0020] Where w is the weighting coefficient, qi is the joint angle of the hip, knee, and ankle joint modules to be solved in this step, and qi1 is the joint angle of the hip, knee, and ankle joint modules in the previous step.

[0021] v(qi)=[Fx(qi),Fy(qi),Fz(qi),τx(qi),τy(qi),τz(qi)]=T(qi)[frs,fls,frf,flf]T is the net external force and net external torque of the center of mass containing the desired joint angles of the hip, knee, and ankle joint modules. T(qi) is the transformation matrix from thrust to the net external force and net external torque of the robot. [frs,fls,frf,flf]T is the thrust of the current two-degree-of-freedom shoulder and foot modules of the robot. qd is the preset joint angle of the robot's hip, knee, and ankle joint modules. a1 and a2 are both positive real numbers and are the coefficients of the corresponding penalty terms.

[0022] Furthermore, in step S4, the constraint inequalities Qu, Ql, loss function L2, and rotation information of the hip joint module, knee joint module, and ankle joint module are solved by a non-optimization solver to obtain the rotation information of the target hip joint module, knee joint module, and ankle joint module.

[0023] Furthermore, in step S5, the rotation angle information of the target hip joint module, knee joint module, and ankle joint module is linearly discretized. The specific formula is shown below: q t =t×(q) i -q i-1 ) / dT+q i-1 ;

[0024] Where qt is the discrete value q i The expected joint angle at the current moment, q i-1 This represents the desired joint angle at the previous moment.

[0025] Furthermore, the present invention also provides a thrust control method for a flying humanoid robot, characterized by comprising the following steps:

[0026] S1: Obtain the position and attitude information of the flying humanoid robot, and the expected state value sequence x of the robot's center of mass over several future periods. r The actual state value x corresponding to the current moment i ;

[0027] S2: Based on the position and attitude information of the flying humanoid robot, the Newton-Euler method is used to establish the spatial dynamics equations, and the spatial dynamics equations are linearly discretized to obtain the discretized spatial dynamics equations, namely the single-particle spatial dynamics equations:

[0028] The superscript o indicates that the item is described in the world frame, the superscript B indicates that the item is described in the centroid coordinate system, and the superscript P indicates that the item is described in the orientation body coordinate system. The origin of the orientation body coordinate system coincides with the origin of the centroid coordinate system and the direction is the same as that of the world frame. E3 is a 3×3 identity matrix, I is the robot's inertia tensor, p is the robot's centroid velocity, ω is the robot's centroid angular velocity, F is the net external force on the robot's centroid, and τ is the net external torque on the robot's centroid.

[0029] S2: Obtain the sequence of expected state values ​​of the robot's centroid over several future periods. r The actual state value x corresponding to the current moment i ;

[0030] S3: Define the constant matrices Q and R of the optimization function, and define the constraint inequality diag(f) for the output. Tmin ...f Tmin )≤U≤diag(f Tmax ...f Tmax ), f Tmin f Tmax The minimum and maximum thrust values ​​for the defined two-DOF shoulder and foot modules are used to construct the loss function L1 = X. T QX+U T RU, where Q matrix is ​​the state variable weight matrix and R matrix is ​​the output variable weight matrix;

[0031] S4: Based on the discretized spatial dynamics equations, the constant matrices Q and R of the optimization function, and the sequence of expected state values ​​of the robot's center of mass over several future periods, x... r The actual state value x corresponding to the current moment i And the constraint inequalities of the output quantity, through a non-optimization solver, yield the desired two-degree-of-freedom shoulder module thrust f. rs ,f ls Represented as a three-dimensional vector, it includes the direction and magnitude of the thrust, and the desired foot module thrust f. rf f lf It is represented as a one-dimensional scalar, describing only the magnitude of the thrust;

[0032] S5: Amplitude limiting is applied based on the actual thrust of the two-degree-of-freedom shoulder and foot modules.

[0033] ||f j ||=min(||fj ||,f Tmax ), thus obtaining the thrust of the two-degree-of-freedom shoulder module after amplitude limiting, fT min fT max The minimum and maximum values ​​of the thrust of the two-degree-of-freedom shoulder module and the thrust of the foot module are set respectively; then the target angles of the first and second joints of the two-degree-of-freedom shoulder module are calculated.

[0034] Where f rs f ls For the desired two-degree-of-freedom shoulder module thrust, f rf f lf For the desired foot module thrust, f Tmax This represents the maximum thrust of the two-degree-of-freedom shoulder module and the maximum thrust of the foot module.

[0035] S6: Based on the limited-range two-degree-of-freedom shoulder module and the desired foot module thrust f rf f lf The target angles of the first and second joints of the two-degree-of-freedom shoulder module control the thrust of the joints of the two-degree-of-freedom shoulder module and the foot module.

[0036] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0037] This invention changes the joint angles and thus the thruster layout based on the current thrust of the flying robot's fan, the rotation angles of the hip, knee, and ankle joints, and the expected and actual state values ​​of the center of mass. This makes the thrust direction as oriented as possible towards the direction of movement, thereby improving flight performance without changing the number and power of the robot's existing thrusters, thus overcoming the limitations of the thrusters themselves. Attached Figure Description

[0038] Figure 1 This is a structural diagram of a flying humanoid robot as shown in Example 1;

[0039] Figure 2 This is a structural diagram of the upper frame shown in Example 2.

[0040] Figure 3 Structural diagram of the leg structure shown in Embodiment 2

[0041] Figure 4 This is a flowchart of a joint deformation control method for a flying humanoid robot, as shown in Example 3.

[0042] In the diagram: 1: Waist base; 2: Upper body frame; 3: Power supply module; 4: Control device; 5: Two-degree-of-freedom shoulder module; 501: Shoulder joint first degree-of-freedom module; 502: Shoulder joint second degree-of-freedom module; 6: Leg structure; 601: Hip joint module; 602: Thigh module; 603: Thigh module; 604: Foot module; 605: Knee joint module; 606: Ankle joint module; 7: First propulsion module; 8: Second propulsion module; 9: Drive motor; 10: First motor; 11: First motor frame; 12: Second motor; 13: Second motor frame; 14: Connector; 15: First link structure plate; 16: Second link structure plate; 17: Third connecting structure. Detailed Implementation

[0043] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent.

[0044] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0045] Example 1:

[0046] This embodiment provides, as follows: Figure 1 The humanoid flying robot shown is characterized by comprising a waist base 1, an upper body frame 2, a power module 3, and a control device 4. The upper body frame 2 is connected to the waist base 1, and the power module 3 and control device 4 are mounted on the upper body frame 2. The upper body frame 2 has symmetrical and identical two-degree-of-freedom shoulder modules 5 on both sides. Two leg structures 6 are symmetrically connected to both sides of the waist base. The leg structure 6 is composed of four modules connected in sequence: a hip joint module 601, a thigh module 602, a lower leg module 603, and a foot module 604. The thigh module 602 and the lower leg module 603 are hinged by a knee joint module 605, and the lower leg module 603 and the foot 604 are hinged by an ankle joint module 606. A first propulsion module 7 is fixed in place. The second propulsion module 8 is fixed to the back of the dual-degree-of-freedom shoulder module 5 and is fixed to the side of the foot module 604, moving together with the foot module 604. A drive motor 9 is installed on the waist base 1, and the output end of the drive motor 9 is connected to the upper frame 2. The drive motor 9 drives the relative rotation between the upper frame 2 and the waist base 1. The power module 3 supplies power to the control device 4 and the drive devices of the dual-degree-of-freedom shoulder module 5, hip joint module 601, thigh module 602, calf module 603, and foot module 604 respectively. The control device 4 is electrically connected to the dual-degree-of-freedom shoulder module 5, hip joint module 601, thigh module 602, calf module 603, foot module 604, knee joint module 605, and ankle joint module 606 respectively.

[0047] During flight, the control device controls the first propulsion module 7 and its thrust. Then, based on the current state of the robot, the angles of the three joint modules, namely the hip joint module 601, the knee joint module 605, and the ankle joint module 606, are adjusted to change the joint rotation angle and thus change the propulsion layout. This makes the thrust direction as oriented as possible towards the direction of movement. In other words, without changing the number and power of the robot's existing propulsion, the flight performance can be further improved, thereby overcoming the limitations of the propulsion itself and improving flight performance.

[0048] Example 2:

[0049] This embodiment continues to disclose the following based on Embodiment 1: Figure 2 The dual-degree-of-freedom shoulder module 5 of the flying humanoid robot shown includes a first-degree-of-freedom shoulder joint module 501 and a second-degree-of-freedom shoulder joint module 502. The drive mechanism for the first-degree-of-freedom shoulder joint module 501 includes a first motor 10, a first motor frame 11, and a first propulsion module 7. The first motor 10 is fixed to the upper frame 2 via the first motor frame 11. The drive mechanism for the second-degree-of-freedom shoulder joint module 502 includes a second motor 12, a second motor frame 13, and the first propulsion module 7. The first propulsion module contains two thrusters, located at the output end of the second motor 12 and the back side, respectively, and are fixed together by a connector 14 to perform tumbling motion. The output end of the second motor 12 is connected to the second motor frame 13, which drives the second-degree-of-freedom shoulder joint module 502, thus enabling the second-degree-of-freedom shoulder joint module 502 to achieve pitch motion. This allows the control device of the flying robot to control the flying humanoid robot to achieve pitch motion.

[0050] And this implementation, for example Figure 3 The lower leg module 603 shown is equipped with a first link structure plate 15 and a second link structure plate 16. A third connecting structure 17 is provided between the first link structure plate 15 and the second link structure plate 16 to enhance the structural rigidity of the lower leg. An electric speed controller (ESC) is placed on the link structure plates 15 and 16 to drive the second propulsion module 8 located on the side of the foot module 604. Four tensioning wheels of the ankle joint drive module are respectively arranged on both sides of the first link structure plate 15, the second link structure plate 16, and the third connecting structure 17. The ankle joint module 606 is driven by a synchronous belt. The passive end of the ankle joint module 606 is connected to the foot module 604. Deep groove ball bearings and needle roller bearings are used to allow the foot module 604 and the lower leg module 603 to rotate relative to each other. This enhances the structural rigidity of the lower leg module, the foot module, and the ankle joint module, providing durability to the joints of the flying humanoid robot when frequently rotating its ankle joint module.

[0051] The output end of the second motor 13 is directly connected to the fan mounting component on one side of the first propulsion module. The connector also has mounting holes, which can be used to connect the robotic arm to form a complete robot arm.

[0052] A fixing plate is provided on the rear side of the first motor 11. The fixing plate is fixed to the other side of the upper frame square tube, surrounds the side wall of the first motor 11, and together with the first motor frame 12, forms a simply supported beam structure to assist in fixing the motor.

[0053] The power module 3 is mounted on the back of the robot's upper frame, providing power to the thrusters, joint drive modules, and control board. The control device 4 is hinged to the front of the upper frame. A groove for mounting straps is provided on the upper side of the control device 4's base plate, allowing the control board base plate to rotate and unfold downwards. There are four ESCs 11, corresponding to the four thrusters of the first propulsion module, and they are fixed to the corresponding thruster side via connectors.

[0054] The hip joint module includes three motors. The first motor is fixed to the waist base and controls the yaw rotation of the thigh module. The second motor is fixed to the motor base and controls the tumbling rotation of the thigh module. Its base is connected to the output end of the first motor. The third motor controls the pitch rotation of the thigh module. Its base is connected to the output end of the second motor. The thigh assembly is connected to the output end of the third motor through side plates. The first, second, and third motors are perpendicular to each other in direction, and their output axes intersect at a point.

[0055] The drive motor of the knee joint drive module is fixed to the linkage structure plate of the thigh. The output end is connected to the synchronous pulley of the active end of the knee joint drive module. The tension wheel mounting bracket is fixed to the drive motor, and the two tension wheels of the knee joint drive module are mounted on it.

[0056] The knee joint module uses a synchronous belt for drive. The knee joint rotation axis includes the driven end of the knee joint drive module, the output end of the ankle joint drive module, and the connecting rod structure plate of the thigh module. Multiple deep groove ball bearings and needle roller bearings ensure that the rotation of these three parts is independent. The ankle joint module's drive motor is mounted on the lower leg connecting rod structure plate, with its output end directly connected to a synchronous pulley, which is bolted to the drive motor. A bearing end cap with a deep groove ball bearing is nested within the active synchronous pulley of the ankle joint drive module to isolate the relative movement between the output synchronous pulley of the ankle joint drive module and the thigh connecting rod structure. A bearing with a flange is installed at the lower leg connecting rod structure plate, and the active synchronous pulley of the ankle joint drive module has a nut groove in its inner cavity, using shoulder bolts for connection and support. A semi-circular structural component connects the connecting rod structure plates of the lower leg module to improve structural rigidity. A bearing with a flange is installed at the lower leg connecting rod structure plate. The inner cavity of the active synchronous pulley of the ankle joint drive module has a nut groove. The connecting rod structure plate, needle roller bearing, thigh connecting rod structure plate, deep groove ball bearing, bearing end cap, deep groove ball bearing and synchronous pulley are connected in sequence using shoulder bolts to form the side support.

[0057] During flight, the control device controls the first propulsion module 7 and its thrust. Then, based on the current state of the robot, the angles of the three joint modules, namely the hip joint module 601, the knee joint module 605, and the ankle joint module 606, are adjusted to change the joint rotation angle and thus change the propulsion layout. This makes the thrust direction as oriented as possible towards the direction of movement. In other words, without changing the number and power of the robot's existing propulsion, the flight performance can be further improved, thereby overcoming the limitations of the propulsion itself and improving flight performance.

[0058] Example 3:

[0059] This embodiment provides, as follows: Figure 4 The method for controlling joint deformation of a flying humanoid robot, as shown, includes the following steps:

[0060] S1: Obtain the rotation angle information of the shoulder module, hip joint module, knee joint module, and ankle joint module of the flying humanoid robot at the current moment; obtain the thrust magnitude of the shoulder module and foot module and the expected and actual state values ​​of the center of mass of the flying humanoid robot;

[0061] S2: Obtain the weighting coefficients based on the expected state value and the actual state value of the current centroid;

[0062] S3: Based on the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, set the constraint inequality Q for the output. u Q lFurthermore, based on the thrust magnitude of the two-degree-of-freedom shoulder module and foot module at the current moment, as well as the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, and the weight coefficients, a loss function L2 is constructed.

[0063] S4: Based on the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, and constraint inequality Q u Q l And the L2 loss function is used to obtain the rotation information of the target hip joint module, knee joint module, and ankle joint module;

[0064] S5: Perform linear discretization on the target hip joint module, knee joint module, and ankle joint module, and use the corresponding discrete value selected at the current moment as the rotation angle information of the hip joint module, knee joint module, and ankle joint module in the new state;

[0065] S6: Transmit the rotation angle information of the new state hip joint module, knee joint module, and ankle joint module to the control device to control the hip joint module, knee joint module, and ankle joint module.

[0066] This embodiment uses the rotation angle information of the two-DOF shoulder module, hip joint module, knee joint module, and ankle joint module of the flying humanoid robot at the current moment, the thrust magnitude of the two-DOF shoulder module and foot module, and the expected and actual state values ​​of the center of mass of the flying humanoid robot to construct a loss function and constraints to solve for the rotation angle information of the hip joint module, knee joint module, and ankle joint module in the new state. This information is then transmitted to the control device to change the joint rotation angle of the flying humanoid robot, thereby changing the thruster layout so that the thrust direction is as far as possible towards the direction of movement. In other words, it is possible to further improve the flight performance without changing the number and power of the robot's existing thrusters, thereby breaking through the limitations of the thrusters themselves and improving the flight performance.

[0067] Example 4:

[0068] This embodiment continues to disclose the following content based on Embodiment 3:

[0069] In step S2, based on the expected and actual state values ​​of the robot's center of mass, the current center of mass pose error, current center of mass velocity error, and current center of mass angular velocity error are calculated. These errors are then input into the PD controller in the flying humanoid robot control device to calculate weighting coefficients. The specific formula is shown below:

[0070] w=P((p,Θ)-(p des ,Θ des ))+D((v,ω)-(v des ,ω des ))

[0071] Where w is the weight vector, P and D are diagonal matrices, with the diagonal elements representing the PD coefficients of the corresponding terms, p is the position of the robot's center of mass in the world frame, Θ is the Euler angle of the robot's center of mass in the world frame, v is the velocity of the robot's center of mass in the world frame, and ω is the angular velocity of the robot's center of mass. des Let ω be the expected velocity of the robot's center of mass in the world frame. des Let be the expected value of the body angular velocity of the robot's center of mass.

[0072] The specific formula for the constraint inequality of the output quantity set in step S3 is as follows:

[0073] Q u =min(q) max q i-1 +ωq max ·△T), Q l =max(q) min q i-1 -ω qmax ·△T), q max q min These represent the maximum and minimum rotation angles of the hip, knee, and ankle joint modules, respectively, determined by the robot's structure. ω qmax ΔT represents the upper limit speed of the joint, and ΔT represents the period of the controller output.

[0074] The specific formula for constructing the loss function L2 in step S3 is as follows:

[0075] L2 = -w T v(q i )+a1||q i-1 -q i ||2+a2||q i -q d ||2

[0076] Where w is the weighting coefficient, q i For the joint angles of the hip, knee, and ankle joint modules to be solved this time, q i-1 The joint angles of the hip, knee, and ankle joint modules at the previous moment.

[0077] v(q i )=[F x (q i ), F y (q i ), F z (q i ), τ x (q i ), τ y (q i), τ z (q i )]=T(q i )[f rs f ls f rf f lf ] T Let T(q) be the net external force and net external torque at the center of mass of the desired hip, knee, and ankle joint angles. i ) is the transformation matrix from thrust to the resultant external force and resultant external torque of the robot, [f rs f ls f rf f lf ] T For the thrust of the current robot's two-DOF shoulder and foot modules, q d The preset joint angles for the robot's hip, knee, and ankle joint modules are a1 and a2, which are both positive real numbers and are the coefficients of the corresponding penalty terms.

[0078] In step S4, the constraint inequality Q is... u Q l The loss function L2 and the rotation information of the hip joint module, knee joint module, and ankle joint module are solved by a non-optimization solver to obtain the rotation information of the target hip joint module, knee joint module, and ankle joint module.

[0079] In step S5, the rotation information of the target hip joint module, knee joint module, and ankle joint module is linearly discretized. The specific formula is shown below: q t =t×(q) i -q i-1 ) / dT+q i-1 ;

[0080] Where q t For discrete values ​​q i The expected joint angle at the current moment, q i-1 This represents the desired joint angle at the previous moment.

[0081] This embodiment uses the rotation angle information of the two-DOF shoulder module, hip joint module, knee joint module, and ankle joint module of the flying humanoid robot at the current moment, the thrust magnitude of the two-DOF shoulder module and foot module, and the expected and actual state values ​​of the center of mass of the flying humanoid robot to construct a loss function and constraints to solve for the rotation angle information of the hip joint module, knee joint module, and ankle joint module in the new state. This information is then transmitted to the control device to change the joint rotation angle of the flying humanoid robot, thereby changing the thruster layout so that the thrust direction is as far as possible towards the direction of movement. In other words, it is possible to further improve the flight performance without changing the number and power of the robot's existing thrusters, thereby breaking through the limitations of the thrusters themselves and improving the flight performance.

[0082] Example 5:

[0083] This embodiment provides, as follows: Figure 3 The thrust control method for a flying humanoid robot shown includes the following steps:

[0084] S1: Obtain the position and attitude information of the flying humanoid robot, and the expected state value sequence x of the robot's center of mass over several future periods. r The actual state value x corresponding to the current moment i ;

[0085] S2: Based on the position and attitude information of the flying humanoid robot, the Newton-Euler method is used to establish the spatial dynamics equations, and the spatial dynamics equations are linearly discretized to obtain the discretized spatial dynamics equations, namely the single-particle spatial dynamics equations:

[0086] The superscript o indicates that the item is described in the world frame, the superscript B indicates that the item is described in the centroid coordinate system, and the superscript P indicates that the item is described in the orientation body coordinate system. The origin of the orientation body coordinate system coincides with the origin of the centroid coordinate system and the direction is the same as that of the world frame. E3 is a 3×3 identity matrix, I is the robot's inertia tensor, p is the robot's centroid velocity, ω is the robot's centroid angular velocity, F is the net external force on the robot's centroid, and τ is the net external torque on the robot's centroid.

[0087] S2: Obtain the sequence of expected state values ​​of the robot's centroid over several future periods. r The actual state value x corresponding to the current moment i ;

[0088] S3: Define the constant matrices Q and R of the optimization function, and define the constraint inequality diag(f) for the output. Tmin ...f Tmin )≤U≤diag(f Tmax ...f Tmax ), f Tmin f TmaxThe minimum and maximum thrust values ​​for the defined two-DOF shoulder and foot modules are used to construct the loss function L1 = X. T QX+U T RU, where Q matrix is ​​the state variable weight matrix and R matrix is ​​the output variable weight matrix;

[0089] S4: Based on the discretized spatial dynamics equations, the constant matrices Q and R of the optimization function, and the sequence of expected state values ​​of the robot's center of mass over several future periods, x... r The actual state value x corresponding to the current moment i And the constraint inequalities of the output quantity, through a non-optimization solver, yield the desired two-degree-of-freedom shoulder module thrust f. rs f ls Represented as a three-dimensional vector, it includes the direction and magnitude of the thrust, and the desired foot module thrust f. rf f lf It is represented as a one-dimensional scalar, describing only the magnitude of the thrust;

[0090] S5: Amplitude limiting is applied based on the actual thrust of the two-degree-of-freedom shoulder and foot modules.

[0091] ||f j ||=min(||f j ||,f Tmax ), thus obtaining the thrust of the two-degree-of-freedom shoulder module after amplitude limiting, f Tmin f Tmax The minimum and maximum values ​​of the thrust of the two-degree-of-freedom shoulder module and the thrust of the foot module are set respectively; then the target angles of the first and second joints of the two-degree-of-freedom shoulder module are calculated. Where f rs f ls For the desired two-degree-of-freedom shoulder module thrust, f rf f lf For the desired foot module thrust, f Tmax This represents the maximum thrust of the two-degree-of-freedom shoulder module and the maximum thrust of the foot module.

[0092] S6: Based on the limited-range two-degree-of-freedom shoulder module and the desired foot module thrust f rf f lf The target angles of the first and second joints of the two-degree-of-freedom shoulder module control the thrust of the joints of the two-degree-of-freedom shoulder module and the foot module.

[0093] This embodiment obtains the position and attitude information of the flying humanoid robot and the expected state value sequence x of the robot's center of mass over several future periods. r The actual state value x corresponding to the current moment iThe desired thrust of the two-degree-of-freedom shoulder module and foot module, as well as the target angles of the first and second joints of the two-degree-of-freedom shoulder module, are obtained. Furthermore, the two-degree-of-freedom shoulder module is subjected to secondary amplitude limiting so that the thrust of the flying robot can be increased while meeting safety constraints, thereby improving flight performance.

[0094] In summary, this invention provides a flying humanoid robot and its joint deformation control method and thrust control method. The flying humanoid robot of this invention includes a waist base, an upper frame, a power module, and a control device. The control device stores the joint deformation control method and the thrust control method. This method can change the joint rotation angle according to the current state of the robot, thereby changing the thruster layout so that the thrust direction is as close as possible to the direction of movement. That is, it can further improve the flight performance without changing the number and power of the robot's existing thrusters, thereby overcoming the inherent properties of the thrusters and improving the flight performance.

[0095] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for controlling joint deformation in a flying humanoid robot, characterized in that, Includes the following steps S1: Obtain the rotation angle information of the shoulder module, hip joint module, knee joint module, and ankle joint module of the flying humanoid robot at the current moment; obtain the thrust magnitude of the shoulder module and foot module and the expected and actual state values ​​of the center of mass of the flying humanoid robot; S2: Obtain the weighting coefficients based on the expected state value and the actual state value of the current centroid; S3: Based on the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, set the constraint inequality Q for the output. u Q l Furthermore, based on the thrust magnitude of the two-degree-of-freedom shoulder module and foot module at the current moment, as well as the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, and the weight coefficients, a loss function L2 is constructed. S4: Based on the rotation information of the two-degree-of-freedom shoulder module, hip joint module, knee joint module, and ankle joint module, and constraint inequality Q u Q l And the L2 loss function is used to obtain the rotation information of the target hip joint module, knee joint module, and ankle joint module; S5: Perform linear discretization on the target hip joint module, knee joint module, and ankle joint module, and use the corresponding discrete value selected at the current moment as the rotation angle information of the hip joint module, knee joint module, and ankle joint module in the new state; S6: Transmit the rotation angle information of the new state hip joint module, knee joint module, and ankle joint module to the control device to control the hip joint module, knee joint module, and ankle joint module; The flying humanoid robot includes a waist base (1), an upper body frame (2), a power module (3), and a control device (4). The upper body frame (2) is connected to the waist base (1). The power module (3) and the control device (4) are mounted on the upper body frame (2). The upper body frame (2) has symmetrical and identical two-degree-of-freedom shoulder modules (5) on both sides. The waist base has two symmetrically connected leg structures (6) on both sides. The leg structures (6) consist of four modules. The modules are connected in the following order: hip joint module (601), thigh module (602), calf module (603), and foot module (604). The thigh module (602) and the calf module (603) are hinged by a knee joint module (605), and the calf module (603) and the foot module (604) are hinged by an ankle joint module (606). The first propulsion module (7) is fixed to the dual-degree-of-freedom shoulder module (5). On the upper body frame (2), the second propulsion module (8) is fixed to the side of the foot module (604) and moves together with the foot module (604). The waist base (1) is equipped with a drive motor (9). The output end of the drive motor (9) is connected to the upper body frame (2). The drive motor (9) drives the upper body frame (2) to rotate relative to the waist base (1). The power module (3) supplies power to the control device (4) and the drive devices of the dual-degree-of-freedom shoulder module (5), the hip joint module (601), the thigh module (602), the calf module (603), and the foot module (604). The control device (4) is electrically connected to the dual-degree-of-freedom shoulder module (5), the hip joint module (601), the thigh module (602), the calf module (603), the foot module (604), the knee joint module (605), and the ankle joint module (606).

2. The method for controlling joint deformation of a flying humanoid robot according to claim 1, characterized in that, The dual-degree-of-freedom shoulder module (5) includes a first degree-of-freedom shoulder joint module (501) and a second degree-of-freedom shoulder joint module (502). The driving device for the first degree-of-freedom shoulder joint module (501) includes a first motor (10) and a first motor frame (11). The first motor (10) is fixed to the upper frame (2) through the first motor frame (11). The driving device for the second degree-of-freedom shoulder joint module (502) includes a second motor (12), a second motor frame (13), and a first propulsion module (7). The first propulsion module contains two propellers, located at the output end of the second motor (12) and the back side, respectively. They are fixed together by a connector (14) to perform tumbling motion. The output end of the second motor (12) is connected to the second motor frame (13) that drives the second degree-of-freedom shoulder joint module (502) to achieve pitching motion.

3. The method for controlling joint deformation of a flying humanoid robot according to claim 1, characterized in that, The lower leg module (603) is provided with a first connecting rod structure plate (15) and a second connecting rod structure plate (16). A third connecting structure (17) is provided between the first connecting rod structure plate (15) and the second connecting rod structure plate (16) to strengthen the structural rigidity of the lower leg. An electric speed controller is placed on the first connecting rod structure plate (15) and the second connecting rod structure plate (16) to drive the second propulsion module (8) located on the side of the foot module (604). Four tensioning wheels of the ankle joint drive module are respectively arranged on both sides of the first connecting rod structure plate (15), the second connecting rod structure plate (16) and the third connecting structure (17). The ankle joint module (606) is driven by a synchronous belt. The passive end of the ankle joint module (606) is connected to the foot module. A deep groove ball bearing and a needle roller bearing are used to enable the foot module (604) and the lower leg module (603) to rotate relative to each other.

4. The method for controlling joint deformation of a flying humanoid robot according to claim 1, characterized in that, In step S2, based on the expected and actual state values ​​of the robot's center of mass, the current center of mass pose error, current center of mass velocity error, and current center of mass angular velocity error are calculated. These errors are then input into the PD controller in the flying humanoid robot control device to calculate weighting coefficients. The specific formula is shown below: w=P((p,Θ)-(p des ,I des ))+D((v,ω)-(v des ,oh des )) Where w is the weight vector, P and D are diagonal matrices, with the diagonal elements representing the PD coefficients of the corresponding terms, p is the position of the robot's center of mass in the world frame, Θ is the Euler angle of the robot's center of mass in the world frame, v is the velocity of the robot's center of mass in the world frame, and ω is the angular velocity of the robot's center of mass. des Let ω be the expected velocity of the robot's center of mass in the world frame. des Let be the expected value of the body angular velocity of the robot's center of mass.

5. The method for controlling joint deformation of a flying humanoid robot according to claim 1, characterized in that, The specific formula for the constraint inequality of the output quantity set in step S3 is as follows: Q u =min(q) max ,q i-1 +ω qmax ·ΔT),Q l =max(q) min ,q i-1 -ω qmax ·ΔT),q max q min These represent the maximum and minimum rotation angles of the hip, knee, and ankle joint modules, respectively, determined by the robot's structure. ω qmax ΔT represents the upper limit speed of the joint, and ΔT represents the period of the controller output.

6. The method for controlling joint deformation of a flying humanoid robot according to claim 4, characterized in that, The specific formula for constructing the loss function L2 in step S3 is as follows: L2=-w T v(q i )+a1||q i-1 -q i ||2+a2||q i -q d ||2 Where w is the weighting coefficient, q i For the joint angles of the hip, knee, and ankle joint modules to be solved this time, q i-1 The joint angles of the hip, knee, and ankle joint modules at the previous moment. v(q i )=[F x (q i ),F y (q i ),F z (q i ),τ x (q i ),τ y (q i ),τ z (q i )]=T(q i )[f rs ,f ls ,f rf ,f lf ] T Let T(q) be the net external force and net external torque at the center of mass of the desired hip, knee, and ankle joint angles. i ) is the transformation matrix from thrust to the resultant external force and resultant external torque of the robot, [f rs ,f lls ,f rf ,f lf ] T For the thrust of the current robot's two-DOF shoulder and foot modules, q d The preset joint angles for the robot's hip, knee, and ankle joint modules are a1 and a2, which are both positive real numbers and are the coefficients of the corresponding penalty terms.

7. The method for controlling joint deformation of a flying humanoid robot according to claim 1, characterized in that, In step S4, the constraint inequality Q is... u Q l The loss function L2 and the rotation information of the hip joint module, knee joint module, and ankle joint module are solved by the optimization solver to obtain the rotation information of the target hip joint module, knee joint module, and ankle joint module.

8. The method for controlling joint deformation of a flying humanoid robot according to claim 1, characterized in that, In step S5, the rotation information of the target hip joint module, knee joint module, and ankle joint module is linearly discretized. The specific formula is shown below: q t =t×(q) i -q i-1 ) / dT+q i-1 ; Where q t For discrete values ​​q i The expected joint angle at the current moment, q i-1 This represents the desired joint angle at the previous moment.

Images