Joint mechanical arm trajectory tracking control method of rehabilitation robot

Abstract

The invention discloses a trajectory tracking control method for a joint mechanical arm of a rehabilitation robot, relates to the technical field of robot control, and adopts an extreme learning machine and combines a position, speed and moment hybrid control method of a virtual control variable. A joint mechanical arm agent model with the extreme learning machine introduced has the high learning speed and good generalization performance, and when the problems of nonlinearity, parameter time-varying characteristics, target variability and the like of complex motion tracks of a rehabilitation robot are solved, the position-speed-torque mixed control method is adopted, and high dynamic performance is achieved. Compared with a model prediction control method, an extreme learning machine is adopted and combined with a position, speed and torque mixed control method of virtual control variables, the online optimization solution process is fast, the parameter identification error precision is high, the rehabilitation robot joint mechanical arm posture compensation operation is more flexible, the complexity and uncertainty of a system model are reduced, and the system performance is improved. And the method has relatively high robustness, relatively high response speed and multi-working-condition adaptability.

Description

Technical Field

[0001] The present invention relates to the technical field of robot control, and particularly to a trajectory tracking control method for the joint manipulator of a rehabilitation robot. Background Art

[0002] In the joint manipulator of a rehabilitation robot, the joint is equivalent to a mass-spring-damping system, where the spring stiffness coefficient, damping coefficient, and mass coefficient are generally determined by the robot automatically sensing and identifying to adjust its body shape and movement gait when contacting different terrains to adapt to a specific terrain environment. Currently, the trajectory tracking control methods for rehabilitation robots mainly include model predictive control methods, feedforward neural networks, iterative learning control, adaptive fuzzy sliding mode observers, etc.

[0003] After retrieving relevant patents and literature at home and abroad, regarding the trajectory tracking control method for the joint manipulator of a rehabilitation robot, for example, the paper "Research on Compliant Control of a Series-Parallel Hybrid Upper Limb Rehabilitation Robot" designed a control method. First, the Lagrangian method was used to perform dynamic modeling on the rehabilitation robot, then the force interaction relationship between humans and machines was established based on the admittance principle, and then a sliding mode control was designed on the basis of dynamics to track the desired trajectory and a fuzzy control was introduced to reduce the chattering of the sliding mode control. The smaller the training intensity of this method, the better the compliance of the control, but there are also problems such as parameter time-variation and target variability. Another paper, "Compliant Control of a Horizontal Rehabilitation Robot Based on an Impedance Model", adopted a compliant control method based on an impedance model, regarded the interaction between humans and robots as a virtual impedance model, and conducted research on the compliant control of the rehabilitation robot based on this impedance model, realizing the change of the motion trajectory according to the human intention, but the parameter adjustment of this method is cumbersome. Another example is the patent application with the application number 202410371420.4, "Active Compliant Control Method for an Ankle Rehabilitation Robot Based on Human-Machine Interaction Torque", which combines the motion deviation to update the desired speed in real time and then updates the speed control data for the electric push rod, and then drives the electric push rod to perform corresponding actions to drive the ankle rehabilitation robot to move, assisting the subject to perform active rehabilitation exercises according to the set rehabilitation motion trajectory and reducing the motion error. This method has a high requirement for the accuracy of the feedforward compensation modeling, increasing the control difficulty.

[0004] Although there are studies on the control methods of rehabilitation robots in the prior art, there are still deficiencies when facing problems such as the nonlinearity, parameter time-variation, and target variability of the complex motion trajectory of the joint manipulator. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a trajectory tracking control method for the joint manipulator of a rehabilitation robot. By using an extreme learning machine and combining the position, speed, and torque hybrid control method of the joint manipulator of the rehabilitation robot, the online optimization solution process is relatively fast, the parameter identification error accuracy is high, the compensation for the attitude operation of the joint manipulator of the rehabilitation robot is more compliant, the complexity and uncertainty of the control system model of the joint manipulator of the rehabilitation robot are reduced, and it has strong robustness, fast response speed, and multi-condition adaptability.

[0006] The present invention adopts the following technical solutions to solve the above technical problems:

[0007] In the first aspect, a trajectory tracking control method for the joint manipulator of a rehabilitation robot proposed by the present invention includes:

[0008] Establish a dynamic model of the joint manipulator of the rehabilitation robot, and obtain the interaction torque of the rehabilitation robot by establishing the dynamic model;

[0009] Define the angular error of the joint manipulator in the Cartesian space according to the interaction torque;

[0010] Substitute the angular error into the angular compensation controller of the joint manipulator of the rehabilitation robot;

[0011] Design a virtual control variable based on the angular compensation controller;

[0012] According to the relationship between the end force and torque generated by the joint manipulator of the rehabilitation robot, considering the virtual control variable, transform the Cartesian space compensation force vector into the torque in the joint space;

[0013] On the basis of the torque in the joint space, set a saturation function to constrain the torque in the joint space, introduce the joint manipulator proxy model of the extreme learning machine, obtain the joint target torque, and realize trajectory tracking control according to the joint target torque.

[0014] As a further optimized solution of the trajectory tracking control method for the joint manipulator of a rehabilitation robot described in the present invention, establish a dynamic model of the joint manipulator of the rehabilitation robot, and obtain the interaction torque θ of the rehabilitation robot d : Specifically as follows:

[0015]

[0016] Wherein, M d 、H d 、I d are the inertia matrix, damping matrix, and stiffness matrix respectively, e f is the trajectory tracking error of the joint manipulator, is e fThe first derivative of represents the velocity error, is the second derivative of e f and represents the acceleration error.

[0017] As a further optimization scheme of the trajectory tracking control method for the joint manipulator of a rehabilitation robot according to the present invention, according to the interaction torque θ d , the angular error ε(t) of the joint manipulator in the Cartesian space is defined;

[0018]

[0019] where is the desired interaction torque, and K c is the coefficient between the length and torque of the joint manipulator, and t is time.

[0020] As a further optimization scheme of the trajectory tracking control method for the joint manipulator of a rehabilitation robot according to the present invention, the angular error ε(t) is substituted into the angular compensation controller of the joint manipulator of the rehabilitation robot

[0021] where K p , K i , K d respectively represent the proportional term, integral term, and differential term parameters of the angular compensation controller, and

[0022] is the angular velocity error. As a further optimization scheme of the trajectory tracking control method for the joint manipulator of a rehabilitation robot according to the present invention, a virtual control variable η k+i is designed based on the angular compensation controller ;

[0023]

[0024] where γ x represents the position error variable of the joint manipulator, γ y represents the velocity error variable of the joint manipulator, γ z represents the torque error variable of the joint manipulator, represents the estimated position of the joint manipulator at the k + i-th moment, is the actual position of the joint manipulator at the k + i-th moment, is the estimated velocity of the joint manipulator at the k + i-th moment, is the actual velocity of the joint manipulator at the k + i-th moment, is the estimated torque of the joint manipulator at the k + i-th moment, is the actual torque of the robotic arm at the k + i moment, where i, k, and N are all integers.

[0025] As a further optimization of the trajectory tracking control method for the robotic arm of a rehabilitation robot described in the present invention, according to the relationship between the end force and torque generated by the robotic arm of the rehabilitation robot, considering the virtual control variable, the Cartesian space compensation force vector is transformed into the torque τ c (t);

[0026] τ c (t) = α c J T (q)η k+i (5)

[0027] where α c is the Cartesian space compensation force vector, J T (q) is the kernel matrix, and q is the joint angle.

[0028] As a further optimization of the trajectory tracking control method for the robotic arm of a rehabilitation robot described in the present invention, based on the torque in the joint space, a saturation function K(γ x , γ y , γ z ) is set to constrain τ c (t), and a surrogate model of the robotic arm of the extreme learning machine is introduced to obtain the joint target torque According to the joint target torque trajectory tracking control is realized;

[0029]

[0030] where is the joint target torque, φ d (t) is the kernel function, d(t) is the lumped disturbance term, and K(γ x , γ y , γ z ) is a saturation function with respect to γ x , γ y , γ z , and K(γ x , γ y , γ z ) is used to compensate for the torque angle error of the robotic arm.

[0031] As a further optimization of the trajectory tracking control method for the robotic arm of a rehabilitation robot described in the present invention,

[0032] where δ is the kernel parameter and exp(·) is the exponential function.

[0033] A computer device includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, it implements the steps of the joint manipulator trajectory tracking control method of the rehabilitation robot according to the first aspect or any corresponding embodiment described above.

[0034] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, it implements the steps of the joint manipulator trajectory tracking control method of the rehabilitation robot according to the first aspect or any corresponding embodiment described above.

[0035] Compared with the prior art, the present invention adopts the above technical solutions and has the following technical effects:

[0036] (1) Compared with the traditional feedforward neural network, the joint manipulator proxy model introducing the extreme learning machine has a faster learning speed and good generalization performance. In the face of problems such as the nonlinearity, parameter time-variation, and target variability of the complex motion trajectory of the rehabilitation robot, the position-velocity-torque hybrid control method adopted has a higher control accuracy.

[0037] (2) Compared with the model predictive control method, the present method adopts the extreme learning machine and combines the position, velocity, and torque hybrid control method of virtual control variables, with a faster online optimization solution process, a high accuracy of parameter identification error, a more compliant compensation for the attitude operation of the joint manipulator of the rehabilitation robot, reducing the complexity and uncertainty of the joint manipulator control system model of the rehabilitation robot, and having strong robustness, a fast response speed, and multi-condition adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flowchart of the present invention;

[0039] Figure 2 is a graph of experimental results of position tracking;

[0040] Figure 3 is a graph of experimental results of speed tracking. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0042] The present invention adopts the extreme learning machine and combines the position, velocity, and torque hybrid control method of virtual control variables to implement a joint manipulator trajectory tracking control method for a rehabilitation robot, reducing the complexity and uncertainty of the joint manipulator control system model of the rehabilitation robot, and having strong robustness, a fast response speed, and multi-condition adaptability.

[0043] A trajectory tracking control method for the joint manipulator of a rehabilitation robot, which is realized by combining the extreme learning machine with the goal of hybrid control of the position, speed and torque of the virtual control variable, includes the following steps:

[0044] First, considering the electrical structure characteristics of the joint manipulator and the fact that it often operates at low speeds, the dynamic model of the joint manipulator of the rehabilitation robot is completed. Establishing the dynamic model enables the rehabilitation robot to obtain the interaction so that the joint manipulator can better achieve the compliance and safety of trajectory tracking control. Through the transformation of the desired position in the Cartesian space to the desired joint angle position in the joint space, according to the hybrid control method of the position, speed and torque of the joint manipulator, the position error variable, speed error variable and torque error variable are set to make the joint manipulator of the rehabilitation robot flexible and adjustable. According to the relationship between the end force of the joint manipulator and the joint torque, the force in the Cartesian space is transformed into the torque in the joint space. To avoid the actual human-machine interaction torque angle exceeding the safe motion range trajectory, an agent model of the joint manipulator of the extreme learning machine is introduced, and a saturation function constraint torque regarding the position error variable, speed error variable and torque error variable is given to compensate for the joint manipulator torque angle error, realizing the output constraint performance of the joint target torque and the trajectory tracking accuracy. It can be seen from the experimental results that introducing the agent model of the joint manipulator of the extreme learning machine can well achieve position and speed tracking.

[0045] The Newton-Euler method is a classical method for establishing the dynamic equation of a robot. Considering the electrical structure characteristics of the joint manipulator and the fact that it often operates at low speeds, the dynamic model of the joint manipulator of the rehabilitation robot is established according to the Newton-Euler equation, and the interaction torque θ of the rehabilitation robot is obtained by establishing the dynamic model d ;

[0046]

[0047] where M d 、H d 、I d are the inertia matrix, damping matrix and stiffness matrix respectively, e f is the trajectory tracking error of the joint manipulator, is the first derivative of e f , represents the speed error, is the second derivative of e f , represents the acceleration error. By establishing the dynamic model, the rehabilitation robot can obtain the interaction so that the joint manipulator can better achieve the compliance and safety of trajectory tracking control.

[0048] According to the interaction torque θd Define the angular error ε(t) of the Cartesian space joint manipulator to improve the controller accuracy of the Cartesian space trajectory tracking performance.

[0049]

[0050] Among them, is the desired interaction torque, and K c is the coefficient between the length and torque of the joint manipulator, and t is time.

[0051] Substitute the angular error ε(t) into the angular compensation controller of the joint manipulator of the rehabilitation robot.

[0052] Among them, K p , K i , K d respectively represent the proportional, integral, and differential parameter terms of the angular compensation controller. is the angular velocity error.

[0053] The excellent performance of the joint manipulator controller is the premise for realizing high-precision Cartesian space trajectory tracking control. To improve the tracking performance, design the virtual control variable η k+i ,

[0054]

[0055] Among them, γ x represents the position error variable of the joint manipulator, γ y represents the velocity error variable of the joint manipulator, γ z represents the torque error variable of the joint manipulator. represents the estimated position of the joint manipulator at the k + i moment. is the actual position of the joint manipulator at the k + i moment. is the estimated velocity of the joint manipulator at the k + i moment. is the actual velocity of the joint manipulator at the k + i moment. is the estimated torque of the joint manipulator at the k + i moment. is the actual torque of the joint manipulator at the k + i moment. i, k, and N are all integers. Among them, k is a fixed value, and i is a variable value. By setting the position error variable, velocity error variable, and torque error variable, the joint manipulator of the rehabilitation robot can be flexibly adjusted.

[0056] According to the relationship between the end force and torque generated by the joint manipulator of the rehabilitation robot, considering the virtual control variable, transform the Cartesian space compensation force vector into the torque τ c (t)

[0057] τ c (t) = α c J T (q)η k+i (5)

[0058] Among them, α c is the Cartesian space compensation force vector, J T (q) is the kernel matrix, and q is the joint angle.

[0059] To avoid the actual human-machine interaction torque angle exceeding the safe motion range trajectory, a joint manipulator proxy model of the extreme learning machine is introduced, and a saturation function K(γ x , γ y , γ z ) about the position error variable, velocity error variable, and torque error variable is set to constrain τ c (t), thereby compensating for the joint manipulator torque angle error and achieving the output constraint performance and trajectory tracking accuracy of the joint target torque .

[0060]

[0061] Among them, is the joint target torque, φ d (t) is the kernel function, d(t) is the lumped disturbance term, and K(γ x , γ y , γ z ) is a saturation function about γ x , γ y , γ z ; K(γ x , γ y , γ z ) is used to compensate for the torque angle error of the joint manipulator;

[0062]

[0063] Among them, δ is the kernel parameter, and exp(·) is the exponential function.

[0064] From the experimental results such as Figure 2 position tracking and Figure 3 velocity tracking, it can be seen from the results in Figure 2 , 3 that introducing the joint manipulator proxy model of the extreme learning machine can well achieve position and velocity tracking.

[0065] This method adopts an extreme learning machine and combines a position, speed, and torque hybrid control method with virtual control variables to achieve a trajectory tracking control method for the joint manipulator of a rehabilitation robot. Compared with traditional feedforward neural networks, the joint manipulator proxy model incorporating the extreme learning machine has a faster learning speed and good generalization performance. When facing problems such as the nonlinearity, parameter time-variation, and target variability of the complex motion trajectory of the rehabilitation robot, the position-speed-torque hybrid control method has high dynamic performance. Compared with the model predictive control method, the method of adopting an extreme learning machine and combining the position, speed, and torque hybrid control method with virtual control variables has a faster online optimization solution process, high accuracy of parameter identification error, more compliant attitude operation for compensating the joint manipulator of the rehabilitation robot, reduces the complexity and uncertainty of the system model, and has strong robustness, fast response speed, and multi-condition adaptability.

[0066] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, it implements the steps of the trajectory tracking control method for the joint manipulator of the rehabilitation robot as described in the first aspect or any corresponding implementation manner above.

[0067] An embodiment of the present invention also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps of the trajectory tracking control method for the joint manipulator of the rehabilitation robot as described in the first aspect or any corresponding implementation manner above.

[0068] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes. The solutions in the embodiments of the present invention can be implemented in various computer languages, for example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0069] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and combinations of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general purpose computers, special purpose computers, embedded processors, or other programmable data processing devices to produce a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices create means for implementing the functions specified in the flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in multiple blocks.

[0070] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction means that implements the functions specified in the flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0071] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0072] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0073] Obviously, those skilled in the art can make various changes and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A joint mechanical arm trajectory tracking control method for a rehabilitation robot, characterized in that: include: Establish a dynamic model of the joint mechanical arm of the rehabilitation robot, and obtain the interactive torque of the rehabilitation robot by establishing the dynamic model; According to the interaction torque, the angular error of the Cartesian space joint robot is defined; Substituting the angle error into the angle compensation controller of the joint manipulator of the rehabilitation robot; Design virtual control variables based on angle compensation controller; According to the relationship between the end force and torque generated by the joint manipulator of the rehabilitation robot, the virtual control variable is considered to transform the Cartesian space compensation force vector into the torque in the joint space; Based on the torque in the joint space, a saturation function is set to constrain the torque in the joint space. The joint robot agent model of the extreme learning machine is introduced to obtain the joint target torque, and trajectory tracking control is achieved according to the joint target torque.

2. The trajectory tracking control method of a joint mechanical arm of a rehabilitation robot according to claim 1, characterized in that: Establish the dynamic model of the joint manipulator of the rehabilitation robot, and obtain the interaction torque θ of the rehabilitation robot by establishing the dynamic model d :Specific details are as follows: Among them, M d , H d ,I d are the inertia matrix, damping matrix and stiffness matrix respectively, e f is the joint robot trajectory tracking error, Yes f The first derivative of represents the speed error, Yes f The second-order derivative of Indicates the acceleration error.

3. The trajectory tracking control method of a joint mechanical arm of a rehabilitation robot according to claim 2, characterized in that: According to the interaction moment θ d , define the angular error ε(t) of the Cartesian space joint robot; in, is the expected interaction torque, K c is the coefficient between the length and torque of the joint robot, and t is the time.

4. The trajectory tracking control method of a joint mechanical arm of a rehabilitation robot according to claim 3, characterized in that: Substitute the angle error ε(t) into the angle compensation controller of the joint manipulator of the rehabilitation robot Among them, K p , K i , K d Respectively represent the proportional term, integral term and differential term parameters of the angle compensation controller, is the angular velocity error.

5. The trajectory tracking control method of a joint mechanical arm of a rehabilitation robot according to claim 4, characterized in that: Angle compensation controller Design dummy control variable η k+i ; Among them, γ x represents the position error variable of the joint robot arm, γ y represents the velocity error variable of the joint robot, γ z represents the torque error variable of the joint robot, represents the estimated position of the joint manipulator at time k+i, is the actual position of the joint robot at time k+i, is the estimated velocity of the joint manipulator at time k+i, is the actual speed of the joint robot at time k+i, is the estimated torque of the joint manipulator at time k+i, is the actual torque of the joint robot at time k+i, where i, k and N are all integers.

6. The trajectory tracking control method of a joint mechanical arm of a rehabilitation robot according to claim 5, characterized in that: According to the relationship between the end force and torque generated by the joint manipulator of the rehabilitation robot, the Cartesian space compensation force vector is transformed into the torque τ in the joint space by taking into account the virtual control variable. c (t); t c (t)=a c J T (q)n k+i (5) Among them, α c is the Cartesian space compensation force vector, J T (q) is the kernel matrix, and q is the joint angle.

7. The trajectory tracking control method of a joint mechanical arm of a rehabilitation robot according to claim 6, characterized in that: Based on the moment in the joint space, the saturation function K(γ x , γ y , γ z ) Constraint τ c (t), introduce the joint robot agent model of the extreme learning machine to obtain the joint target torque According to the joint target torque Realize trajectory tracking control; in, is the joint target torque, φ d (t) is the kernel function, d(t) is the lumped disturbance term, K(γ x , γ y , γ z ) is about γ x , γ y , γ z The saturation function, K(γ x , γ y , γ z ) is used to compensate for the torque angle error of the joint robot.

8. The trajectory tracking control method of a joint mechanical arm of a rehabilitation robot according to claim 7, characterized in that: Among them, δ is the kernel parameter and exp(·) is the exponential function.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: When the processor executes the computer program, the steps of the joint mechanical arm trajectory tracking control method of the rehabilitation robot as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the joint mechanical arm trajectory tracking control method of a rehabilitation robot as described in any one of claims 1 to 8 are implemented.

Images