Industrial robot adaptive impedance control method, device and medium based on state transfer algorithm

Abstract

The present invention discloses an industrial robot adaptive impedance control method, device and medium based on a state transfer algorithm, the method comprising: constructing a dynamic model of an industrial robot based on the Newton-Euler method, and identifying unknown parameters of the model by an experimental identification method; designing a position impedance controller for the industrial robot in contact with the external environment based on the dynamic model obtained by identification, and introducing an adaptive control strategy to ensure the stability of the force error; optimizing the parameters of the adaptive impedance controller using a state transfer algorithm, and introducing a variable scale mechanism to dynamically adaptively update the optimization parameters to speed up the optimization speed; controlling the industrial robot based on an adaptive impedance controller with optimized parameters. The present invention realizes the compliant control of the end of the industrial robot under the premise of fully considering the dynamic characteristics of the industrial robot, can improve the robustness of the robot in an unknown environment, and correspondingly optimizes the control performance.

Description

Technical Field

[0001] The present invention belongs to the field of robot joint control, and relates to a six-degree-of-freedom industrial robot, and in particular to an industrial robot adaptive impedance control method, device and medium based on a state transfer algorithm. Background Art

[0002] Industrial robots are usually used in flexible assembly, surface polishing and other work scenarios. In these work scenarios, it is necessary not only to track the running trajectory, but also to control the force of the end effector. Impedance control can achieve compliant control of the end force by converting the end force into position deviation through the impedance model. As for the traditional impedance control strategy, although it can achieve compliant control of the end force to a certain extent, for scenarios with unknown environments, the traditional impedance control strategy still has corresponding tracking errors. Therefore, in such task environments with complex and changeable motion environments, the application of traditional impedance control still has certain limitations. Summary of the invention

[0003] In view of the above technical problems, the present invention provides an industrial robot adaptive impedance control method, device and medium based on a state transfer algorithm, which has a better force tracking effect.

[0004] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0005] An industrial robot adaptive impedance control method based on a state transfer algorithm, comprising:

[0006] S1: Construct the dynamic model of the industrial robot based on the Newton-Euler method, and use the experimental identification method to identify the unknown parameters of the dynamic model;

[0007] S2: Based on the identified dynamic model, a position impedance controller is designed for the industrial robot to contact the external environment, and an adaptive control strategy is introduced into the designed position impedance controller to obtain an adaptive impedance controller;

[0008] S3: Optimizing the control parameters of the adaptive impedance controller by using a state transfer algorithm, and introducing a variable scale mechanism to dynamically and adaptively update the optimization parameters in the algorithm;

[0009] S4: An adaptive impedance controller based on the optimized parameters obtained in S3 is used to control the industrial robot.

[0010] Furthermore, the constructed dynamic model takes into account the friction characteristics and is expressed as:

[0011]

[0012] Among them, τ linkFor the constructed kinetic model, q, Corresponding to joint angle, angular velocity and angular acceleration respectively; D r (q) represents the inertial force vector, represents the Coriolis force, centrifugal force and gravity term vector; τ friction represents the friction force vector, and its expression is f c ,f v Respectively represent the Coulomb and viscous friction coefficients, sign() is the sign function; τ act represents the external disturbance torque.

[0013] Furthermore, the unknown parameters of the dynamic model are identified by the experimental identification method as follows:

[0014] The external disturbance torque τ act Zeroing, linearizing the kinetic model and obtaining a minimum uniquely identifiable parameter set;

[0015] Based on the dynamic model of the minimum uniquely identifiable parameter set, Fourier series is used as the excitation trajectory, and the constant coefficients of the excitation trajectory are optimized by genetic algorithm;

[0016] Control the industrial robot to execute the excitation trajectory after constant coefficient optimization, and obtain the actual angular displacement, angular velocity, angular acceleration and torque by collecting the angular displacement and current data of the industrial robot during operation;

[0017] According to the actual angular displacement, angular velocity, angular acceleration and torque, the unknown parameters of the dynamic model are identified.

[0018] Furthermore, the designed position impedance controller is described by the second-order differential equation:

[0019]

[0020] Among them, X, Corresponding to the actual position, velocity, acceleration, X e , Corresponding to the desired position, velocity, and acceleration of the end; F exp ,F act Represents the expected external contact force and the actual contact force; M r ,B r ,K r denote the inertia, damping and stiffness matrices respectively;

[0021] The adaptive impedance controller obtained by introducing the adaptive control strategy is expressed as:

[0022]

[0023] Among them, Ω(t) represents the adaptive compensation of force error, and its expression is T represents the controller sampling period and η represents the update rate.

[0024] Furthermore, the dynamic model is rewritten as:

[0025]

[0026] Among them, the intermediate quantity D r An estimated value of

[0027] Then, by introducing the time delay law into the dynamic model, we can obtain Estimated value of As Substituting into the rewritten dynamic model, the adaptive impedance control rate is obtained as:

[0028]

[0029] in, It can be obtained by taking the second-order derivative of the original dynamic model, and its expression is: J(q) is the force Jacobian matrix.

[0030] Furthermore, the step S3 specifically includes:

[0031] S3-1: Determine the optimization objective function, the maximum search intensity SE and the maximum number of iterations Maxiter according to the analysis;

[0032] S3-2: Determine the feasible range of the inertia parameters, damping parameters and stiffness parameters to be optimized, and randomly generate an initial optimal individual Best in the corresponding feasible search space k ;

[0033] S3-3: Use the scaling operator ET, the rotation operator RT and the coordinate transformation operator AT to generate SE sample individuals to be screened, and use the greedy criterion to update the current optimal solution; when the optimal individual changes, use the translation operator TT to generate new SE sample individuals, and update the optimal individual;

[0034] S3-4: adaptive parameter update, after performing a round of optimization, adaptively update the rotation factor in the rotation operator and the translation factor in the translation operator using a variable scale mechanism;

[0035] S3-5: Determine whether the termination condition of the iteration is reached, and if the maximum number of iterations is not reached, return to step S3-3;

[0036] S3-6: Output the inertia parameters, damping parameters and stiffness parameters corresponding to the optimal individual.

[0037] Furthermore, the objective function in step S3-1 is specifically:

[0038] The quadratic performance index function is selected as the optimization target, that is, the direction of the minimum combined end position error and contact force error is optimized. The designed objective function is:

[0039]

[0040] Among them, m r ,b r ,k r Respectively represent M r ,B r ,K r Inertia parameters, damping parameters and stiffness parameters in e x ,e f They represent the position error and force error of the end of the industrial robot respectively. ρ is the contact force error calculation constant, and its value range is 0<ρ<1.

[0041] Furthermore, the scaling mechanism in step S3-4 is specifically as follows:

[0042] A counter is used to record the relative number of magnitudes count in the optimization process. The calculation formula is:

[0043]

[0044] Among them, μ is the critical value of the difference between the objective functions of adjacent individuals. When the twice calculated objective functions are greater than the critical value, the value of count is increased by one, otherwise it is decreased by one.

[0045] Then the rotation factor α and translation factor β are updated, and the update criteria are as follows:

[0046]

[0047] Among them, V1 and V2 represent the upper and lower critical values ​​of the relative magnitude, and ζ is the scale transformation factor.

[0048] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor implements any one of the above-mentioned methods for adaptive impedance control of industrial robots based on a state transfer algorithm.

[0049] A computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the method for adaptive impedance control of an industrial robot based on a state transfer algorithm described in any one of the above is implemented.

[0050] Beneficial Effects

[0051] In view of the control limitations of the existing impedance control technology in the unknown environment, the present invention introduces an adaptive control algorithm to realize active and compliant control of the end force on the basis of fully considering the dynamic characteristics of the robot, which can improve the robustness of the robot in an unknown environment. At the same time, in order to further obtain better control performance, the present invention uses a state transfer algorithm to optimize the control parameters, and introduces a variable scale mechanism to improve it to accelerate the convergence speed, which can improve the end force control accuracy of industrial robots in complex processes such as polishing and assembly. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 The present invention is a schematic diagram of the implementation flow of an industrial robot adaptive impedance control method based on a state transfer algorithm.

[0053] Figure 2 The present invention is a schematic structural diagram of a serial six-degree-of-freedom industrial robot according to an embodiment of the present invention.

[0054] Figure 3 The present invention relates to a block diagram of an adaptive impedance control based on a dynamic model.

[0055] Figure 4 The present invention is a schematic diagram of the process of optimizing the adaptive impedance parameters of an industrial robot based on a state transfer algorithm.

[0056] Figure 5 This is a schematic diagram of simulation results obtained after optimizing control parameters using three different methods involved in the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the present invention more clear, the technical solution in the embodiment of the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein with reference to the accompanying drawings are only used to explain the present invention and are not used to limit the present invention.

[0058] like Figure 1 As shown, the present invention proposes an industrial robot adaptive impedance control method based on a state transfer algorithm, which specifically includes the following steps:

[0059] Step 1: According to the kinematic characteristics, the industrial robot dynamics model is constructed based on the Newton-Euler method, and the unknown parameters of the model are identified using the experimental identification method.

[0060] The industrial robot used in this embodiment is a serial six-degree-of-freedom industrial robot, and its structural diagram is as follows: Figure 2As shown, the corresponding improved DH parameter coordinates are shown in Table 1.

[0061]

[0062]

[0063] Then, the Newton-Euler method is used to perform recursive analysis based on the above improved DH coordinate parameters. In this process, the friction factor is taken into consideration to obtain the corresponding dynamic model as follows:

[0064]

[0065] The unknown model parameters are obtained by using the experimental identification method, the external force terms are set to zero and the model is linearized, and then the minimum identifiable parameter set of 48 is obtained; the excitation trajectory of the identification process is optimized using a genetic algorithm, and finally the robot is controlled to execute the excitation trajectory, and the joint angular displacement and current data during the operation are collected; the collected current data is preprocessed and then inserted into the identification equation to obtain the unknown parameters of the model.

[0066] Step 2: Design the corresponding position impedance controller according to the dynamic model obtained by identification, and introduce the adaptive control strategy to further design the adaptive impedance controller.

[0067] like Figure 3 As shown, this embodiment is based on position-type impedance control and implements the design of an adaptive impedance controller by introducing the following adaptive control rate.

[0068]

[0069] In Matlab's simulink simulation software, the corresponding controller simulation model is built according to the above control rate combined with the dynamic model obtained by identification.

[0070] Step 3: Using a state transfer algorithm to adjust the impedance parameters of the adaptive impedance controller, while introducing a variable scale mechanism to dynamically and adaptively update the optimal parameters to accelerate the convergence speed;

[0071] In this embodiment, after the controller is built, it is necessary to further optimize the impedance control parameters, including inertia, damping and stiffness coefficients. Figure 4 As shown, the process mainly includes the following specific steps:

[0072] (1) Determine the objective function and initialize the optimization parameters

[0073] In this embodiment, the direction with the minimum combined end position error and contact force error is selected for optimization. The designed objective function is:

[0074]

[0075] where e x ,e f Represent the position error and force error of the end, respectively, ρ is the contact force error calculation constant, and its value is set to 0.5. After determining the objective function, according to the complexity analysis of the system, the maximum search force SE and the maximum number of iterations Maxiter are set to 10 and 50 respectively.

[0076] (2) Determine the optimization range of the parameters and initialize the particles. Here, the upper and lower limits of the particle optimization are set to 0 and 100 respectively, and then an initial optimal individual Best is randomly generated. k .

[0077] (3) The telescopic transformation operator ET, the rotation operator RT and the coordinate transformation operator AT are used to generate SE sample individuals to be screened. At the same time, the greedy criterion is used to update the current optimal solution. When the optimal individual changes, the translation operator TT is used to generate new SE sample individuals and update the optimal individual.

[0078] The telescopic transformation operator ET is specifically:

[0079] Best k+1 =Best k +γR e Best k

[0080] Among them, γ is a positive constant, R e A diagonal matrix that follows a Gaussian distribution. In this example, this parameter is set to 1.

[0081] The rotation operator RT is specifically:

[0082]

[0083] The rotation factor α is a positive constant, which is a random matrix with each element taking values ​​between [-1, 1]. ||·||2 is the vector binorm. In this example, the parameter value range is set to [0.0001, 1].

[0084] The coordinate transformation operator AT is specifically:

[0085] Best k+1 =Best k +δR a Best k

[0086] Where δ is a positive constant, R ais a random diagonal matrix with only one non-zero value at a random position that follows a Gaussian distribution. In this example, this parameter is set to 1.

[0087] The translation operator TT is specifically:

[0088]

[0089] The translation factor β is a positive constant, R t is a random variable whose value range is [0,1]. In this example, the value range is set to [0.1,10].

[0090] (4) Adaptive parameter update: After executing a round of optimization, the rotation and translation factors in the rotation and translation operators are adaptively updated using the variable scale mechanism. This process uses a counter to record the relative magnitude count in the optimization process, and its calculation formula is:

[0091]

[0092] Wherein, μ is the critical value of the difference between adjacent individual objective functions, and in this embodiment, the value is 0.0001. When the calculated objective function twice is greater than the critical value, the value of count is increased by one, otherwise it is decreased by one. Then the rotation factor and translation factor are updated, and the update criteria are as follows:

[0093]

[0094] Among them, α, β represent the rotation factor and translation factor respectively, V1, V2 represent the upper and lower critical values ​​of the relative magnitude respectively, and their values ​​are set to 4 and 9 respectively, and ζ is the scale transformation factor, and its value is set to 2.

[0095] (5) When the termination condition is reached, the final optimization result is output; otherwise, the process returns to (3) for iterative calculation.

[0096] Under the above-mentioned parameter conditions, the final parameter optimization result based on the variable scale state transfer algorithm is M r =1.059,B r =10.25,K r =0.5.

[0097] Step 4: Set the expected tracking contact force to a sinusoidal form and perform simulations based on the controller parameters obtained from different algorithms.

[0098] In order to further verify the superiority of the present invention, the parameters of the designed adaptive impedance controller are optimized by using genetic algorithm and particle swarm algorithm respectively under the same conditions, and then the control effects of the three algorithms are compared. Figure 5As shown in the figure, it can be seen that the controller output performance corresponding to the adaptive impedance control method based on the state transfer algorithm proposed in the present invention is optimal. Under the condition that other experimental conditions are the same, the system tracking effect corresponding to the parameters obtained by the present invention is the best, the overshoot and steady-state error are significantly reduced, and the flexible control of the end force of the industrial robot is better achieved.

[0099] So far, the specific implementation examples have been described in conjunction with the drawings. It should be understood that the various technologies described above can be combined and improved arbitrarily, and they should be considered to be within the protection scope of the present invention without departing from the core content of the present invention.

Claims

1. An industrial robot adaptive impedance control method based on a state transfer algorithm, characterized in that: include: S1: Construct the dynamic model of the industrial robot based on the Newton-Euler method, and use the experimental identification method to identify the unknown parameters of the dynamic model; S2: Based on the identified dynamic model, a position impedance controller is designed for the industrial robot to contact the external environment, and an adaptive control strategy is introduced into the designed position impedance controller to obtain an adaptive impedance controller; The designed position impedance controller is described by the second-order differential equation: Among them, X, Corresponding to the actual position, velocity, acceleration, X e , Corresponding to the desired position, velocity, and acceleration of the end; F exp ,F act Represents the expected external contact force and the actual contact force; M r ,B r ,K r denote the inertia, damping and stiffness matrices respectively; The adaptive impedance controller obtained by introducing the adaptive control strategy is expressed as: Among them, Ω(t) represents the adaptive compensation of force error, and its expression is T represents the controller sampling period, η represents the update rate; S3: Optimizing the control parameters of the adaptive impedance controller by using a state transfer algorithm, and introducing a variable scale mechanism to dynamically and adaptively update the optimization parameters in the algorithm; S4: An adaptive impedance controller based on the optimized parameters obtained in S3 is used to control the industrial robot.

2. The industrial robot adaptive impedance control method according to claim 1, characterized in that: The constructed dynamic model takes into account the friction characteristics and is expressed as: Among them, τ link For the constructed kinetic model, q, Corresponding to joint angle, angular velocity and angular acceleration respectively; D r (q) represents the inertial force vector, represents the Coriolis force, centrifugal force and gravity term vector; τ friction represents the friction force vector, and its expression is f c ,f v Respectively represent the Coulomb and viscous friction coefficients, sign() is the sign function; τ act represents the external disturbance torque.

3. The industrial robot adaptive impedance control method according to claim 2, characterized in that: The experimental identification method is used to identify the unknown parameters of the dynamic model as follows: The external disturbance torque τ act Zeroing, linearizing the kinetic model and obtaining a minimum uniquely identifiable parameter set; Based on the dynamic model of the minimum uniquely identifiable parameter set, Fourier series is used as the excitation trajectory, and the constant coefficients of the excitation trajectory are optimized by genetic algorithm; Control the industrial robot to execute the excitation trajectory after constant coefficient optimization, and obtain the actual angular displacement, angular velocity, angular acceleration and torque by collecting the angular displacement and current data of the industrial robot during operation; According to the actual angular displacement, angular velocity, angular acceleration and torque, the unknown parameters of the dynamic model are identified.

4. The industrial robot adaptive impedance control method according to claim 1, characterized in that: The kinetic model is rewritten as: Among them, the intermediate quantity D r An estimated value of Then, by introducing the time delay law into the dynamic model, we can obtain Estimated value of As Substituting into the rewritten dynamic model, the adaptive impedance control rate is obtained as: in, It can be obtained by taking the second-order derivative of the original dynamic model, and its expression is: J(q) is the force Jacobian matrix.

5. The industrial robot adaptive impedance control method according to claim 1, characterized in that: The step S3 specifically includes: S3-1: Determine the optimization objective function, the maximum search intensity SE and the maximum number of iterations Maxiter according to the analysis; S3-2: Determine the feasible range of the inertia parameters, damping parameters and stiffness parameters to be optimized, and randomly generate an initial optimal individual Best in the corresponding feasible search space. k ; S3-3: Use the scaling operator ET, the rotation operator RT and the coordinate transformation operator AT to generate SE sample individuals to be screened, and use the greedy criterion to update the current optimal solution; when the optimal individual changes, use the translation operator TT to generate new SE sample individuals, and update the optimal individual; S3-4: adaptive parameter update, after performing a round of optimization, adaptively update the rotation factor in the rotation operator and the translation factor in the translation operator using a variable scale mechanism; S3-5: Determine whether the termination condition of the iteration is reached, and if the maximum number of iterations is not reached, return to step S3-3; S3-6: Output the inertia parameters, damping parameters and stiffness parameters corresponding to the optimal individual.

6. The method for adaptive impedance control of an industrial robot according to claim 5, characterized in that: The objective function in step S3-1 is specifically: The quadratic performance index function is selected as the optimization target, that is, the direction of the minimum combined end position error and contact force error is optimized. The designed objective function is: Among them, m r ,b r ,k r Respectively represent M r ,B r ,K r Inertia parameters, damping parameters and stiffness parameters in e x ,e f They represent the position error and force error of the end of the industrial robot respectively. ρ is the contact force error calculation constant, and its value range is 0<ρ<1.

7. The industrial robot adaptive impedance control method according to claim 5, characterized in that: The scaling mechanism in step S3-4 is specifically: A counter is used to record the relative number of magnitudes count in the optimization process. The calculation formula is: Among them, μ is the critical value of the difference between the objective functions of adjacent individuals. When the twice calculated objective functions are greater than the critical value, the value of count is increased by one, otherwise it is decreased by one. Then the rotation factor α and translation factor β are updated, and the update criteria are as follows: Among them, V1 and V2 represent the upper and lower critical values ​​of the relative magnitude, and ζ is the scale transformation factor.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the computer program is executed by the processor, the processor is caused to implement the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images