A compliant control robot adaptive motion planning method and system

Abstract

The application provides a compliant control robot adaptive motion planning method and system, the method adopts admittance control as a compliance algorithm, is more compliant compared with traditional rigid control, and can select different compliance degrees according to the environment; combines inverse kinematics, trajectory interpolation and environmental parameters to construct an optimal reset trajectory under joint space safety area constraints; improves the flexibility of the robot in a complex working environment by perceiving external forces and dynamically adjusting control parameters; and proposes a chassis-robot arm combined kinematics model to solve the problem of limited working space of a traditional robot arm.

Description

Technical Field

[0001] The present invention belongs to the field of robot technology control, and in particular relates to a method and system for adaptive motion planning of a compliant control robot. Background Art

[0002] Robotics has become a key driver of intelligent manufacturing and automated production, and has been widely used in industrial production, medical rehabilitation, and home services. However, traditional robotic motion planning and control methods often face numerous challenges in complex and changing working environments. For example, in tasks such as industrial assembly, medical surgery, and human-robot collaboration, robots must possess high precision, high compliance, and adaptability to cope with environmental changes, external interference, and safe interaction with humans.

[0003] Currently, robotic motion planning and control primarily rely on rigid control methods. While these methods can achieve high-precision trajectory tracking, they are prone to impact, vibration, and even damage to the equipment during contact tasks or dynamic environments, making them difficult to meet compliance and safety requirements. Furthermore, existing adaptive motion planning technologies are often based on preset models or fixed rules, lacking the ability to perceive and respond to dynamic environments in real time, resulting in insufficient robot adaptability in complex scenarios. Summary of the Invention

[0004] In response to the above problems, the present invention proposes an adaptive motion planning method and system based on a compliant control robot. The concept is to dynamically calculate the desired position of the robot arm and generate a motion trajectory based on the perceived external force, combined with a compliant control and path planning method, so that the robot arm can move smoothly and better complete the target task. The present invention aims to solve the following technical problems: (1) Traditional robots use rigid control and have poor compliance; (2) The existing path planning technology has insufficient adaptability and lacks real-time perception of the dynamic environment; (3) The lack of coordination between the motion planning and control modules leads to limited overall system performance.

[0005] In a first aspect, the present invention provides an adaptive motion planning method for a compliant control robot, comprising the following steps:

[0006] S1: Analyze the collected data information, adjust the compliance control parameters and calculate the desired position;

[0007] S2: Generate a preliminary motion trajectory through trajectory interpolation, and use inverse kinematics to calculate the specific joint angles. If no solution is found, a new solution is found using a coordinated operation of the robot arm and chassis. The final trajectory is interpolated to generate the final trajectory.

[0008] S3: Establish an error feedback compensation system: Calculate the required compensation amount in advance based on the system input signal and the known characteristics of the external disturbance.

[0009] S4: Complete the adaptive path planning task of the robot arm according to the total trajectory.

[0010] S5: Collect historical operation data of the robot arm, calculate important evaluation indexes, optimize the trajectory planning algorithm and the compliance control.

[0011] As preferred, the S1 step specifically comprises:

[0012] S11: Use admittance control as the compliance control algorithm, and the basic formula of the robot arm admittance control is as follows:

[0013]

[0014] In the formula, M d , B d , and K d respectively represent the inertia matrix, the damping matrix, and the stiffness matrix; respectively represent the expected acceleration and the actual acceleration; respectively represent the expected speed and the actual speed; p d , p e respectively represent the expected position and the actual position; h d , h e respectively represent the expected elastic force wrench and the actual elastic force wrench in contact with the environment;

[0015] S12: According to the perceived environmental information, select the compliance strategy:

[0016] When |F e |≥F max , start the high compliance mode: reduce K d , and increase B d , which shows a larger resistance;

[0017] When |F e |<F max , start the low compliance mode: increase K d , and reduce B d , which shows a smaller resistance;

[0018] Wherein, F e is the value of the force applied by a person on the end handle of the robot arm detected by the high-precision six-dimensional force sensor, and F max is the set threshold value of the contact force of the robot arm.

[0019] S13: Calculate the expected acceleration, and use the initial values of and p d for the first calculation, and then use the updated and p d for iterative calculation:

[0020]

[0021] where the initial value is the sampling value of the sensor in the static state.

[0022] S14: Set the control period Δt, calculate and update the expected speed using the expected acceleration:

[0023]

[0024] S15: Calculate and update the expected position using the expected speed:

[0025]

[0026] S16: Integrate and twice-integrate in each control period thereafter to update the expected speed and the expected position p d respectively, so that and p d gradually converge until the admittance control equation is satisfied.

[0027] As a preference, the S2 step specifically comprises:

[0028] S21: Plan a trajectory from the known starting point and target point according to the expected position obtained in the previous step, and generate the joint trajectory using quintic polynomial interpolation for the joint position, velocity and acceleration of the starting point and end point.

[0029] S22: For each trajectory point, solve the corresponding joint angle using inverse kinematics:

[0030] For an n-degree-of-freedom robot arm, each joint of the robot arm is a rotational joint, and the robot arm motion equation is:

[0031] 0 T n = 0 T1(θ1) 1 T2(θ2)… n-1 T n (θ n ) (29)

[0032] where 0 T n represents the rotation matrix from the 0th to the nth joint;

[0033] Solve each axis joint variable θ1,…θ n according to the motion equation.

[0034] S23: Record the number of groups of solutions generated as M, and each group of solutions is recorded as:

[0035]

[0036] S24: Filter valid solutions according to the range of joint motion angle:

[0037]

[0038] If any joint is out of range, exclude solution θ (k) ;

[0039] S25: Define collision function C(θ):

[0040]

[0041] If C(θ (k) ) = 1, exclude solution θ (k) ;

[0042] S26: Use weighted Euclidean distance as the objective function to calculate the minimum total energy consumption of each joint motion:

[0043]

[0044] where w i is the joint weight;

[0045] From the remaining valid solution set S = {θ (k)}, select the solution that minimizes the objective function as the optimal solution θ * :

[0046]

[0047] S27: For each trajectory point, the optimal solution θ * is generated by a quintic polynomial interpolation to generate a continuous and smooth joint trajectory:

[0048] θ i (t) = a i t 5 +b i t 4 +c i t 3 +d i t 2 +e i t+f i (15)

[0049] where a i , b i , c i , d i , e i , f i are polynomial coefficients;

[0050] When no solution exists, determine that the target is unreachable, move the chassis:

[0051] Define the workspace sphere radius as r, and the end position deviation vector as Δp:

[0052] Δp = p d - p0 (16)

[0053] where p0 is the initial position.

[0054] When |Δp| > r, trigger the chassis cooperative motion, generate the chassis desired acceleration:

[0055]

[0056] where K p1 is the proportional gain matrix;

[0057] S28: Convert the acceleration to the global coordinate system, and integrate the acceleration to get the velocity command:

[0058]

[0059] When the velocity threshold is reached, maintain a constant speed, and slow down when approaching the target point.

[0060] S29: After the chassis stops moving, update the coordinate system and the target position coordinate p d , and re-plan the trajectory.

[0061] As a preferred, the degree of freedom of the robot arm is 6, and at this time, the step S22 specifically comprises:

[0062] For each trajectory point, use inverse kinematics to solve the corresponding joint angle:

[0063] For a 6-DOF robot arm, and each joint is a rotational joint, the robot arm motion equation is:

[0064]

[0065] where n, o and a are the unit vectors of the new coordinate system, n x , n y and n z are the projection vectors of the n vector in the XYZ coordinate system, o x , o y and o z are the projection vectors of the o vector in the XYZ coordinate system, a x , a y and a z are the projection vectors of the a vector in the XYZ coordinate system; p x , p y and p zis the position coordinate of the new coordinate system origin in the XYZ coordinate system; the first three axis joint variables are solved:

[0066]

[0067]

[0068] wherein, θ i is the rotation angle of the i-th joint, a i is the length of the connecting rod of the i-th joint, d i is the offset of the i-th joint;

[0069] Since the joint angles of the last three axes only affect the orientation of the robot arm coordinate system, the joint variables of the last three axes can be solved by the Euler angle solution.

[0070] The rotation sequence is represented by a sequence of rotation angles around the x-axis, y-axis, and z-axis.

[0071] For a known attitude rotation matrix R:

[0072]

[0073] Solve the corresponding Euler angles:

[0074]

[0075] The relationship between the joint variables of the last three axes is:

[0076]

[0077] As a preferred embodiment, the S3 step specifically includes:

[0078] S31: Adjust the control signal based on feedforward compensation;

[0079] S32: Adjust the control input u PID (t) in real time according to the current error e(t):

[0080]

[0081] S33: Calculate the required compensation in advance according to the system input signal r(t) and the known characteristics of external disturbance d(t):

[0082] u FF (t) = G FF (r(t), d(t)) (20)

[0083] S34: Output according to the compensation:

[0084] u(t) = u​PID (t)+u FF (t) (21)

[0085] As preferred, the S5 step specifically comprises:

[0086] S51: Evaluate the performance of the robot in the current task from the task completion rate, execution efficiency, and safety:

[0087] Task completion rate:

[0088]

[0089] Execution efficiency:

[0090]

[0091] Where T is the task time, τ(t) is the joint torque vector, and α, β are weight coefficients;

[0092] Safety score based on force and collision detection:

[0093]

[0094] Where Fk is the kth contact force, F is the safety threshold, λ is the penalty coefficient, and N is the total number of tasks performed; k safe

[0095] S52: Evaluate the performance of the robot in the current task from the compliance control performance:

[0096] Contact force tracking error:

[0097]

[0098] F actual is the actual force, and F desired is the desired force;

[0099] Dynamic impedance matching degree:

[0100]

[0101] Where Z(s) is the transfer function of the impedance model, Z actual is the actual impedance transfer function, and Z desired is the desired impedance transfer function; is the norm measure of frequency domain error;

[0102] S53: Based on the feedback data, optimize the compliance control through adaptive impedance parameter learning, and optimize the adaptive motion planning algorithm through reinforcement learning to improve the performance of the system;

[0103] ​​S54: update the control parameters, apply the new parameters to the model, and prepare for the next task.

[0104] In a second aspect, the present application provides a compliant control robot adaptive motion planning system, characterized in that it comprises:

[0105] A compliant control module uses an admittance control as a compliance algorithm to iteratively calculate the desired position based on the acquired force information of the human to the robot arm.

[0106] An adaptive path planning module uses a polynomial interpolation method to generate joint trajectories based on the calculated desired position, and combines an inverse kinematics solving method to calculate specific joint angles and output; when there is no solution, a robot and chassis cooperative operation mode is used to approach the target point and complete the task;

[0107] A feedback and error compensation module uses a feedforward compensation-based controller to adjust the control signal. According to the known characteristics of the system input signal and external disturbance, the required compensation amount is calculated in advance.

[0108] An optimization learning module is responsible for evaluating the performance of the robot in the current task, optimizing the path planning module and the compliant control module.

[0109] The present application has the following beneficial effects:

[0110] The present application uses admittance control as a compliance algorithm, which is more compliant than traditional rigid control, and can select different compliance levels according to the different forces applied by humans, and has better human-machine interaction; an adaptive trajectory planning method is proposed, which combines inverse kinematics, trajectory interpolation and environmental parameters to construct the optimal reset trajectory under the joint space safety region constraint; by sensing external forces and dynamically adjusting control parameters, the flexibility of the robot in complex working environments is improved; a chassis-robot joint kinematics model is proposed to solve the problem of limited working space of traditional robot arms. BRIEF DESCRIPTION OF DRAWINGS

[0111] Figure 1 A flowchart of a compliant control robot adaptive motion planning method.

[0112] Figure 2 A structure diagram of a compliant control robot adaptive motion planning system. DETAILED DESCRIPTION

[0113] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0114] In a real-world example, consider a robot consisting of a 6-DOF arm and a chassis. A force sensor is mounted at the end of the arm. The robot system is initialized, and the sensor collects data.

[0115] As attached Figure 1 As shown, a method for adaptive motion planning of a compliant control robot includes the following steps:

[0116] S1: Analyze the collected data information, adjust the compliance control parameters and calculate the desired position;

[0117] Furthermore, step S1 specifically includes:

[0118] S11: Admittance control is used as the compliant control algorithm. The basic formula of the manipulator's admittance control is as follows:

[0119]

[0120] Where M d 、B d , K d represent the inertia matrix, damping matrix and stiffness matrix respectively; They represent the expected acceleration and actual acceleration respectively; represent the expected speed and actual speed respectively; p d 、p e Denote the expected position and actual position respectively; h d 、h e They represent the expected elastic force screw and the actual elastic force screw in contact with the environment respectively;

[0121] S12: Select a compliant strategy based on the perceived environmental information:

[0122] When |F e |≥F max When high compliance mode is activated: reduce K d , increase B d , which is manifested as greater resistance;

[0123] When |F e | <F max When starting low soft mode: increase K d , reduce B d , which is manifested as less resistance;

[0124] where F e is the value of the force exerted by the human on the end of the robot arm detected by the high-precision six-dimensional force sensor, F max is the set threshold value of the contact force of the robot arm.

[0125] S13: Calculate the desired acceleration, the first time and p d Use the initial value, and then use the updated and p d to perform iterative calculations:

[0126]

[0127] where the initial value is the sampling value of the sensor in the static state.

[0128] S14: Set the control period Δt, and use the desired acceleration to calculate and update the desired velocity:

[0129]

[0130] S15: Use the desired velocity to calculate and update the desired position:

[0131]

[0132] S16: Integrate and twice-integrate in each control period thereafter to update the desired velocity and the desired position p d respectively, so that and p d gradually converge until the admittance control equation is satisfied.

[0133] S2: Generate a preliminary motion trajectory by trajectory interpolation, and calculate the specific joint angles using inverse kinematics solution calculation; when there is no solution, a mechanical arm and chassis cooperative operation mode is used to re-solve. Interpolate the final trajectory points to generate the total trajectory.

[0134] Further, step S2 specifically includes:

[0135] S21: Plan a trajectory from the known starting point and target point according to the desired position obtained in the previous step, and use a quintic polynomial interpolation to generate a preliminary joint trajectory for the joint position, velocity and acceleration of the starting point and end point.

[0136] S22: For each trajectory point, use inverse kinematics to solve the corresponding joint angle:

[0137] For the 6-DOF robot arm in this example, each joint of the robot arm is a rotary joint, and the robot arm motion equation is:

[0138]

[0139] Solve the first three axis joint variables:

[0140]

[0141] where, θ i is the rotation angle of the i-th joint, a i is the length of the connecting rod of the i-th joint, d i is the offset of the i-th joint.

[0142] Since the joint angles of the last three axes only affect the orientation of the robot arm coordinate system, the joint variables of the last three axes can be solved by the Euler angle method.

[0143] Use to represent the rotation sequence around the x-axis, y-axis, and z-axis.

[0144] For the known attitude rotation matrix R:

[0145]

[0146] The corresponding Euler angles can be solved:

[0147]

[0148] The relationship between the joint variables of the last three axes is:

[0149]

[0150] S23: Record the number of solutions generated as M, and each solution is recorded as:

[0151]

[0152] S24: According to the range of joint motion angle, filter the effective solution:

[0153]

[0154] where, and are the minimum and maximum values of the set motion angle respectively.

[0155] If any joint is out of range, exclude the solution θ (k) ;

[0156] S25: Define the collision function C(θ):

[0157]

[0158] If C(θ(k) ) = 1, the solution θ is excluded (k) ;

[0159] S26: Calculate the total energy consumption of each joint motion by using the weighted Euclidean distance as the objective function:

[0160]

[0161] where w i is the joint weight, is the current joint angle value of the robot arm;

[0162] From the remaining valid solution set S = {θ (k)}, select the solution that minimizes the objective function as the optimal solution θ * :

[0163]

[0164] S27: Generate a continuous and smooth joint trajectory for each trajectory point optimal solution θ * :

[0165] θ i (t) = a i t 5 +b i t 4 +c i t 3 +d i t 2 +e i t+f i (15)

[0166] where a i , b i , c i , d i , e i , f i are the polynomial coefficients of the i-th joint;

[0167] When there is no solution, determine that the target is unreachable, and move the chassis:

[0168] Define the workspace sphere radius as r, and the end position deviation vector Δp:

[0169] Δp = p d -p0 (16)

[0170] where p0 is the initial position.

[0171] When |Δp| > r, trigger the chassis cooperative motion to generate the chassis expected acceleration

[0172]

[0173] where K p1 is a proportional gain matrix;

[0174] S28: Convert acceleration to global coordinate system and integrate acceleration to get velocity command:

[0175]

[0176] When the speed threshold is reached, maintain constant speed Slow down when approaching the target point;

[0177] S29: Update coordinate system and target position coordinates p d after the chassis stops moving, re-plan trajectory.

[0178] S3: Establish error feedback compensation system: according to the known characteristics of system input signal and external disturbance, calculate the required compensation in advance.

[0179] Further, step S3 specifically includes:

[0180] S31: Adjust the control signal using a PID controller based on feedforward compensation;

[0181] S32: The PID controller adjusts the control input u PID (t) in real time according to the current error e(t):

[0182]

[0183] S33: The feedforward controller calculates the required compensation in advance according to the system input signal r(t) and the known characteristics of external disturbance d(t):

[0184] u FF (t) = G FF (r(t), d(t)) (20)

[0185] S34: The controller outputs according to the compensation:

[0186] u(t) = u PID (t) + u FF (t) (21)

[0187] S4: Complete the adaptive path planning task of the robot arm according to the total trajectory.

[0188] S41: Drive the motor to execute joint angle and chassis movement.

[0189] S42: When the following conditions are met, the system enters the task end state.

[0190] Mechanical arm reaches the designated position:

[0191] |p e -p d |≤1mm (22)

[0192] Timeout:

[0193] t current -t start >60s (23)

[0194] If other abnormal situations such as collision occur, it will also be stopped urgently into the task end state.

[0195] S43: Reset the control system to the initial state, prepare for the next task.

[0196] S5: Collect historical operation data of the mechanical arm, calculate important evaluation indexes, optimize trajectory planning algorithm and compliant control.

[0197] Further, step S5 specifically includes:

[0198] S51: Evaluate the performance of the robot in the current task from the task completion rate, execution efficiency, and safety:

[0199] Task completion rate:

[0200]

[0201] Execution efficiency:

[0202]

[0203] Where T is the task time, τ(t) is the joint torque vector, and α, β are weight coefficients;

[0204] Safety score based on force and collision detection:

[0205]

[0206] Where F k is the kth contact force, F safe is the safety threshold, and λ is the penalty coefficient;

[0207] S52: Evaluate the performance of the robot in the current task from the compliant control performance:

[0208] Contact force tracking error:

[0209]

[0210] Dynamic impedance matching degree:

[0211]

[0212] wherein Z(s) is the transfer function of the impedance model, is the norm of the frequency domain error;

[0213] S53: According to the feedback data, the adaptive impedance parameter learning is used to optimize the compliance control, and the reinforcement learning is used to optimize the adaptive motion planning algorithm, so as to improve the performance of the system;

[0214] S54: Update the control parameters, apply the new parameters to the model, and prepare for the next task.

[0215] As shown in the accompanying Figure 2 , an adaptive motion planning system for a compliant control robot, characterized in that it comprises:

[0216] A compliant control module uses an admittance control as a compliance algorithm to iteratively calculate the desired position based on the force information obtained from the human operator.

[0217] An adaptive path planning module uses a polynomial interpolation method to generate joint trajectories based on the calculated desired position, and then uses an inverse kinematics solving method to calculate the specific joint angles and output them; when there is no solution, a collaborative operation mode of the robot and the chassis is adopted to approach the target point and complete the task.

[0218] A feedback and error compensation module uses a controller based on feedforward compensation to adjust the control signal. According to the known characteristics of the system input signal and external disturbance, the required compensation amount is calculated in advance.

[0219] An optimization learning module is responsible for evaluating the performance of the robot in the current task, optimizing the path planning module and the compliant control module.

[0220] The above is only a preferred embodiment of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solution that falls within the scope of the present application is within the protection scope of the present application. It should be noted that for ordinary technical personnel in the technical field, some improvements and refinements without departing from the principles of the present application are also considered within the protection scope of the present application.

Claims

1. A compliant control robot adaptive motion planning method, comprising the following steps: S1: analyzing collected data information, adjusting compliant control parameters and calculating expected positions; S2: generating a preliminary motion trajectory through trajectory interpolation, calculating specific joint angles using inverse kinematics solution, and re-solving when there is no solution by using a way of collaborative operation of the robot and the chassis; and generating a total trajectory by interpolating the finally obtained trajectory points; S3: establishing an error feedback compensation system: calculating the required compensation in advance according to the known characteristics of system input signals and external disturbances; S4: completing the adaptive path planning task of the robot according to the total trajectory; S5: collecting historical operation data of the robot, calculating important evaluation indexes, optimizing the trajectory planning algorithm and compliant control; Step S2 comprises: S21: planning a trajectory from a known starting point and target point according to the expected positions obtained in the previous step, and generating a joint trajectory by using a quintic polynomial interpolation for the joint positions, velocities and accelerations of the starting point and the end point; S22: for each trajectory point, solving the corresponding joint angle using inverse kinematics; For a For a 6-DOF manipulator, each joint of the manipulator is a revolute joint, and the motion equation of the manipulator is wherein R0to ndenotes the rotation matrix from the 0thto the nthjoint; Solving the equation of motion to obtain each axis joint variable ; S23: record the number of generated solution groups as Each solution group is recorded as: S24: screening valid solutions according to the motion angle range of the joint; If any joint is out of bounds, then exclude the solution ; S25: Defining a collision function : If = 1, then the solution is excluded ; S26: using weighted Euclidean distance as the objective function to calculate the total energy consumption of minimizing the motion of each joint: wherein, are joint weights; from the remaining active solution set selecting the solution that minimizes the objective function as the optimal solution : S27: Optimal solution for each trajectory point Generate continuous, smooth joint trajectories by quintic polynomial interpolation: wherein are polynomial coefficients; When there is no solution, it is determined that the target is unreachable, and the chassis is moved: The workspace sphere radius is defined as , the end position deviation vector : wherein is the initial position; When a chassis cooperative motion is triggered, resulting in a desired chassis acceleration: wherein is a proportional gain matrix; S28: converting acceleration to the global coordinate system and integrating acceleration to obtain the velocity command: When the velocity threshold is reached, keep constant speed, and slow down after approaching the target point; S29: Update coordinate system and target position coordinates after the chassis stops moving , re-plan trajectory; In S22, the degree of freedom of the robot is 6, and the S22 step specifically comprises: for each trajectory point, solving the corresponding joint angle using inverse kinematics; Each joint of the robot is a rotary joint, and the robot motion equation is: wherein is a unit vector of the new coordinate system, , and are projection vectors of the vector in the XYZ coordinate system, and are projection vectors of the vector in the XYZ coordinate system, , are projection vectors of the vector in the XYZ coordinate system; , and are the position coordinates of the origin of the new coordinate system in the XYZ coordinate system; the first three axis joint variables are solved. wherein, , is the rotation angle of the jth joint, is the link length of the jth joint, is the offset of the jth joint, ​​​ with corner list indicates the sequence of rotations around axes, axes, axes; For known pose rotation matrix : Solve the corresponding Euler angles: The relationship between the joint variables of the last three axes is: 。 2. The method of claim 1, wherein, The S1 step specifically comprises: S11: using admittance control as the compliant control algorithm, and the basic formula of the robot admittance control is as follows: wherein , , respectively represent an inertia matrix, a damping matrix and a stiffness matrix; respectively represent a desired acceleration and an actual acceleration; respectively represent a desired velocity and an actual velocity; respectively represent a desired position and an actual position; respectively represent a desired elastic force wrench and an actual elastic force wrench in contact with the environment; S12: selecting a compliant strategy according to the sensed environmental information: When the high compliance mode is activated: the reduction , the increase , is manifested as a greater resistance; When low compliance mode is initiated: increase , decrease , which is manifested as less resistance; wherein, is the value of the force applied by the person on the mechanical arm end handle detected by the high-precision six-dimensional force sensor, is the set threshold value of the mechanical arm contact force; S13: Calculate expected acceleration, first time and with initial values, then use updated and Iterative calculation: S14: Set control period Calculate and update desired speed using desired acceleration: S15: calculating and updating the expected position using the expected velocity: S16: In each subsequent control cycle Perform integration and quadratic integration to update the expected speed respectively and desired location ,make and Gradually converge until the admittance governing equation is satisfied.

3. The method of claim 1, wherein, The S3 step specifically comprises: S31: adjusting the control signal based on feedforward compensation; S32: Adjusting the control input according to the current error Real-time adjustment of control input : S33: Calculate the required compensation based on the system input signal and the known characteristics of the external disturbance in advance. S34: outputting the compensation amount: 。 4. The method of claim 1, wherein, The S5 step specifically comprises: S51: evaluating the performance of the robot in the current task from the task completion rate, execution efficiency and safety: Task completion rate: Execution efficiency: wherein, is a task duration, is a joint torque vector, is a weight coefficient; Safety score based on force and collision detection: wherein, is the kth contact force, is a safety threshold, is a penalty coefficient, N is the total number of tasks performed; S52: evaluating the performance of the compliant control of the robot in the current task: Contact force tracking error: is the actual force, is the desired force; Dynamic impedance matching degree: wherein, is a transfer function of an impedance model, is an actual impedance transfer function, is a desired impedance transfer function; is a normed measure of frequency domain error; S53: optimizing the compliant control through adaptive impedance parameter learning based on feedback data, and optimizing the adaptive motion planning algorithm through reinforcement learning to improve the performance of the system; S54: updating the control parameters, applying the new parameters and the model, and preparing for the next task.

5. An adaptive motion planning system based on compliant control robot, applied to the adaptive motion planning method based on compliant control robot in claim 1, characterized in that, Compliant control module, using an admittance control as a compliant algorithm, iteratively calculating the expected position according to the force information of the human to the robot; ​ Adaptive path planning module, according to the calculated desired position using a polynomial interpolation method to generate joint trajectory, combined with a kind of inverse kinematics solution method to calculate the specific joint angle and output; When there is no solution, a kind of collaborative operation mode of robot and chassis is adopted to complete the task near the target point; Feedback and error compensation module, using a kind of controller based on feedforward compensation to adjust the control signal; According to the known characteristics of system input signal and external disturbance, the required compensation is calculated in advance; Optimization learning module is responsible for evaluating the performance of the robot in the current task, optimizing the path planning module and compliant control module.

Images