Repetitive motion control method and system for mobile robots based on discrete-time neural dynamics
By constructing the kinematic equations of the position and velocity layers of a mobile robot using a discrete-time neurodynamics approach, and combining an exponential penalty strategy and a five-step explicit linear multistep rule, the accuracy and applicability issues of real-time repetitive motion control for mobile robots are solved, achieving high-precision real-time repetitive motion control.
Patent Information
- Application Number
- CN202210713454.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-22
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-06-22
AI Technical Summary
Existing technologies struggle to provide efficient, real-time, repeatable motion control for mobile robots, particularly in terms of computational accuracy and applicability, leading to frequent occurrences of non-repetitive motion phenomena.
By employing a discrete-time neurodynamics-based approach, position-level kinematic equations, velocity-level kinematic equations, and physical limit constraints are constructed. Combined with an exponential penalty strategy and a five-step explicit linear multistep rule, a discrete-time nullified neurodynamics model is built to achieve high-precision real-time repetitive motion control of a mobile robot.
By updating the calculation once within each calculation time interval, the solution result at the next time point is accurately predicted, eliminating non-repetitive motion phenomena, improving motion real-time performance and path tracking accuracy, and effectively handling physical limit constraints.
Smart Images

Figure CN114952860B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of robot control, and more specifically, to a method and system for controlling repetitive motion of a mobile robot based on discrete-time neural dynamics. Background Art
[0002] The mobile robot is mainly composed of a mobile platform and a robotic arm fixed on the mobile platform. The repetitive motion control of the mobile robot requires that after completing the closed path tracking, the final state of the mobile platform and the robotic arm are consistent with the initial state. If the mobile platform and the robotic arm do not return to the initial state, that is, non-repetitive motion occurs, the mobile robot needs to be additionally configured or self-moved, which greatly reduces the efficiency of task execution.
[0003] A high-performance redundant manipulator repetitive motion planning method and device are currently available, including designing a quadratic repetitive motion performance index, generating a quadratic optimization redundancy analytical solution; converting the quadratic optimization redundancy analytical solution of the steps into a quadratic programming form; solving the quadratic programming using a quadratic programming solver and passing the solution result to a lower computer controller to drive the redundant manipulator motion.
[0004] However, the above methods express the repetitive motion control problem of mobile robots as a time-varying quadratic optimization problem and use continuous-time neurodynamics methods or numerical algorithms to solve it. However, there are problems such as low calculation accuracy and difficulty in directly applying to discrete-time systems. The numerical algorithm requires repeated iterative calculations at each calculation time point to obtain the optimal solution, and cannot perform efficient real-time repeatable motion control of mobile robots. Summary of the invention
[0005] In order to overcome the defect in the prior art that a mobile robot cannot be efficiently and real-timely repeated motion controlled, the present invention provides a method and system for repetitive motion control of a mobile robot based on discrete-time neural dynamics.
[0006] In order to solve the above technical problems, the technical solution of the present invention is as follows:
[0007] In a first aspect, the present invention proposes a method for controlling repetitive motion of a mobile robot based on discrete-time neural dynamics, comprising the following steps:
[0008] S1: Collect motion data of mobile robot;
[0009] S2: construct the position layer kinematic equation, velocity layer kinematic equation and physical limit constraints of the mobile robot, and convert the physical limit constraints into double-end inequality constraints of the velocity layer;
[0010] S3: Combining the position layer kinematic equation, velocity layer kinematic equation and velocity layer double-end inequality constraints, a repetitive motion control scheme of a mobile robot based on a time-varying quadratic optimization problem is constructed;
[0011] S4: Based on the zeroing neurodynamics design rule, an exponential penalty strategy is introduced into the repetitive motion control scheme of the mobile robot based on the time-varying quadratic optimization problem to construct a continuous-time zeroing neurodynamics model;
[0012] S5: constructing a discrete-time annulization neural dynamics model based on the five-step explicit linear multi-step rule and the continuous-time annulization neural dynamics model;
[0013] S6: inputting the motion data of the mobile robot into the discrete time annulization neural dynamics model to calculate the control variables of the mobile robot;
[0014] S7: Performing repeated motion control on the mobile robot according to the control variables.
[0015] In a second aspect, the present invention proposes a mobile robot repetitive motion control system based on discrete-time neural dynamics, comprising:
[0016] A collection module, used to collect motion data of the mobile robot;
[0017] The first building module is used to build the position layer kinematic equation, velocity layer kinematic equation and physical limit constraints of the mobile robot, and convert the physical limit constraints into velocity layer double-end inequality constraints;
[0018] The second building module is used to construct a repetitive motion control scheme of the mobile robot based on a time-varying quadratic optimization problem by combining the position layer kinematic equation, the velocity layer kinematic equation and the velocity layer double-end inequality constraint;
[0019] The third building block is used to introduce an exponential penalty strategy into the repetitive motion control scheme of the mobile robot based on the time-varying quadratic optimization problem based on the zeroing neural dynamics design rule, and to build a continuous-time zeroing neural dynamics model;
[0020] A fourth construction module is used to construct a discrete-time annulization neural dynamics model according to the five-step explicit linear multi-step rule and the continuous-time annulization neural dynamics model;
[0021] A calculation module, used for calculating the control variables of the mobile robot according to the motion data of the mobile robot and the discrete time annulization neural dynamics model;
[0022] The control module is used to perform repetitive motion control on the mobile robot according to the control variables.
[0023] In a third aspect, the present invention proposes an electronic device comprising a processor and a memory; the memory is used to store programs; the processor executes the program to implement the steps of any scheme in repetitive motion control of a mobile robot based on discrete-time neural dynamics.
[0024] In a fourth aspect, the present invention proposes a computer-readable storage medium storing a program, wherein the program is executed by a processor to implement the steps of any scheme in repetitive motion control of a mobile robot based on discrete-time neural dynamics.
[0025] Compared with the prior art, the beneficial effects of the technical solution of the present invention are as follows: the present invention designs a repetitive motion control scheme for a mobile robot based on a time-varying quadratic optimization problem by constructing the position layer kinematic equations, velocity layer kinematic equations and velocity layer double-end inequality constraints of the mobile robot, and introduces an exponential penalty strategy and a five-step explicit linear multi-step rule on the basis of the repetitive motion control scheme of the mobile robot to construct a discrete time zeroing neural dynamics model, which can accurately predict the solution result at the next time point in each calculation time interval based on current and previous known data information and through a single update calculation, thereby realizing high-precision real-time repetitive motion control of a mobile robot with physical limit constraints, effectively eliminating the non-repetitive motion phenomenon, and greatly improving the real-time motion, path tracking accuracy and physical limit constraint processing capability of the mobile robot. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 Flowchart of the repetitive motion control method for a mobile robot based on discrete-time neural dynamics.
[0027] Figure 2 This is a simplified model diagram of the mobile robot of the present invention.
[0028] Figure 3 This is a kinematic model diagram of the mobile platform of the present invention.
[0029] Figure 4 It is the motion trajectory of the mobile robot of the present invention and the actual trajectory and expected path diagram of the end effector.
[0030] Figure 5 It is a diagram of the connection points between the mobile platform and the mechanical arm and the heading angle of the mobile platform of the present invention.
[0031] Figure 6 It is the element trajectory diagram of the combined angle vector of the mobile robot of the present invention.
[0032] Figure 7 It is the element trajectory diagram of the combined velocity vector of the mobile robot of the present invention.
[0033] Figure 8 It is an element trajectory diagram of the tracking error vector of the end effector of the mobile robot of the present invention.
[0034] Fig. 9 Architecture diagram of a repetitive motion control system for a mobile robot based on discrete-time neural dynamics. DETAILED DESCRIPTION
[0035] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;
[0036] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.
[0037] Example 1
[0038] See also Figure 1 , this embodiment proposes a mobile robot repetitive motion control method based on discrete time neural dynamics, comprising the following steps:
[0039] S1: Collect motion data of the mobile robot.
[0040] S2: Construct the position layer kinematic equation, velocity layer kinematic equation and physical limit constraints of the mobile robot, and convert the physical limit constraints into velocity layer double-end inequality constraints.
[0041] S3: Combining the position layer kinematic equation, velocity layer kinematic equation and velocity layer double-end inequality constraints, a repetitive motion control scheme for a mobile robot based on a time-varying quadratic optimization problem is constructed.
[0042] In this embodiment, by uniformly converting the physical limit constraints of the combined angle vector and the combined velocity vector of the mobile robot into expressions based on the combined velocity vector, namely, the double-ended inequality constraints of the velocity layer, a repetitive motion control scheme of the mobile robot whose decision variable is the combined velocity vector is constructed, so that the repetitive motion control scheme can be restated as a standard time-varying quadratic optimization problem analyzed at the velocity layer.
[0043] S4: Based on the zeroing neurodynamics design rule, an exponential penalty strategy is introduced into the repetitive motion control scheme of the mobile robot based on the time-varying quadratic optimization problem to construct a continuous-time zeroing neurodynamics model.
[0044] S5: Based on the five-step explicit linear multi-step rule and the continuous-time annihilation neural dynamics model, a discrete-time annihilation neural dynamics model is constructed.
[0045] S6: Input the motion data of the mobile robot into the discrete time annihilation neural dynamics model to calculate the control variables of the mobile robot.
[0046] S7: Performing repeated motion control on the mobile robot according to the control variables.
[0047] By constructing the position layer kinematic equations, velocity layer kinematic equations and velocity layer double-end inequality constraints of the mobile robot, a repetitive motion control scheme for the mobile robot based on the time-varying quadratic optimization problem is designed, and on the basis of the repetitive motion control scheme of the mobile robot, an exponential penalty strategy and a five-step explicit linear multi-step rule are introduced to construct a discrete time zeroing neural dynamics model. In each calculation time interval, it can accurately predict the solution result at the next time point based on the current and previous known data information and through a single update calculation, thereby realizing high-precision real-time repetitive motion control of the mobile robot with physical limit constraints, effectively eliminating the non-repetitive motion phenomenon, and greatly improving the real-time motion, path tracking accuracy and physical limit constraint processing capability of the mobile robot.
[0048] Example 2
[0049] See also Figure 2-Figure 8 , this embodiment proposes a mobile robot repetitive motion control method based on discrete time neural dynamics, comprising the following steps:
[0050] S1: Collect motion data of mobile robot;
[0051] In this embodiment, the motion data of the mobile robot includes the combined angle vector of the mobile robot, the actual position vector of the end effector, the rotation angles of the left drive wheel and the right drive wheel of the mobile platform, the joint angle vector of the six-joint redundant robotic arm fixed on the mobile platform, the heading angle of the mobile platform, the combined velocity vector of the mobile robot, and the expected path of the end effector.
[0052] S2: Construct the position layer kinematic equation, velocity layer kinematic equation and physical limit constraints of the mobile robot, and convert the physical limit constraints into double-end inequality constraints of the velocity layer:
[0053] S2.1: Determine discrete time points of the mobile robot's motion and construct position-layer kinematic equations at the discrete time points.
[0054] In this embodiment, at discrete time points In the above example, the kinematic equation of the position layer is constructed, and its expression is as follows:
[0055] Φ(Θ k+1 )=r actual,k+1
[0056]
[0057] Among them, k is the update index, is the time step, t final is the final time point of the entire calculation time interval, is the nonlinear mapping function array, is the set of real numbers, Θ k+1 is the combined angle vector of the mobile robot, r actual,k+1 is the actual position vector of the end effector, is the rotation angle set of the left and right driving wheels of the mobile platform, θ k+1 is the joint angle vector of the six-joint redundant manipulator fixed on the mobile platform.
[0058] Discrete time points By k and The value is determined, generally refers to the next discrete time point, and t k Generally refers to the current discrete time point. At discrete time point t k+1 The expression can better illustrate that the solution and problem to be solved are for the next discrete time point.
[0059] S2.2: Construct the velocity layer kinematic equation according to the position layer kinematic equation.
[0060] In this embodiment, in order to monitor and control the process of the end effector tracking the desired path, the positioning error vector of the end effector is defined as ∈ k+1 =r actual,k+1 -r desired,k+1 , where r desired,k+1 is the desired path of the end effector.
[0061] By introducing the above-mentioned end effector positioning error as error feedback, the expression of the velocity layer kinematic equation is as follows:
[0062]
[0063]
[0064] Among them, φ k+1 is the heading angle of the mobile platform, is the combined velocity vector of the mobile robot, is the rotation speed set of the left and right driving wheels of the mobile platform, is the joint velocity vector of the six-joint redundant manipulator fixed on the mobile platform, J(φ k+1 ,θ k+1 ) is the matrix associated with Jacobi; ζ k+1 is the vector containing the position error feedback of the end effector, is the time derivative of the desired path of the end effector, r desired,k+1is the expected path of the end effector, and λ1 is a settable parameter.
[0065] S2.3: Construct physical limit constraints by discretizing a set of time-varying two-ended inequalities.
[0066] In this embodiment, the expression of the physical limit constraint of the mobile robot is as follows:
[0067]
[0068] Among them, Θ - represents the lower limit of the combined angle vector of the mobile robot, Θ + represents the upper limit of the combined angle vector of the mobile robot, represents the lower limit of the combined velocity vector of the mobile robot, Represents the upper limit of the combined velocity vector of the mobile robot.
[0069] S2.4: Based on the exponential limit conversion strategy, the physical limit constraint is converted into a velocity layer double-end inequality constraint, the expression of which is as follows:
[0070]
[0071] Among them, α k+1 represents the lower limit of the double-ended inequality constraint, β k+1 represents the upper limit of the double-ended inequality constraint, α k+1 The i-th element α i,k+1 (i=1,2,…,n), and β k+1 The i-th element β i,k+1 (i=1,2,…,n), respectively, are defined as follows:
[0072]
[0073] Where exp(·) represents an exponential function; σ1>0 is a settable parameter used to adjust the deceleration amplitude; parameter σ2∈(0,1) is a settable parameter used to adjust the critical area range that requires deceleration, that is, and
[0074] S3: Combining the position layer kinematic equation, velocity layer kinematic equation and velocity layer double-end inequality constraints, a repetitive motion control scheme for a mobile robot based on a time-varying quadratic optimization problem is constructed:
[0075] S3.1: Combining the position layer kinematic equation, velocity layer kinematic equation and velocity layer double-end inequality constraints, design a repetitive motion control scheme for the mobile robot, which is expressed as follows:
[0076]
[0077]
[0078]
[0079] Among them, D k+1 The coefficient matrix representing the performance index of repeated motion, d k+1 The coefficient vector representing the performance index of repeated motion, D k+1 and d k+1 The expressions are as follows:
[0080]
[0081]
[0082]
[0083] Among them, C k+1 represents the auxiliary matrix related to the heading angle of the mobile platform, λ2, λ3 and λ4 are all positive design parameters; p connection,k+1 =[x connection,k+1 ;y connectioh,k+1 ] is the connection point between the mobile platform and the robotic arm, x connection,k+1 Indicates the coordinate of the connection point on the x-axis, y connection,k+1 represents the coordinate of the connection point on the Y axis; p connection (0) represents the initial state of the connection point between the mobile platform and the robotic arm, φ(0) is the initial state of the heading angle of the mobile platform, θ(0) represents the initial state of the joint angle vector of the robotic arm, a is the distance from the left driving wheel and the right driving wheel to the midpoint of the two axles; b is the distance from the midpoint of the two axles to the connection point p connection,k+1 , γ is the radius of the left driving wheel and the right driving wheel; 0 represents the zero matrix, and I represents the unit matrix.
[0084] S3.2: Convert the repetitive motion control scheme into a time-varying quadratic optimization problem, and obtain a repetitive motion control scheme based on the time-varying quadratic optimization problem, the expression of which is as follows:
[0085]
[0086] sG k+1 x k+1 =g k+1
[0087] H k+1 x k+1 ≤h k+1
[0088] Where T represents the transpose operator of a vector or matrix. is the decision variable vector to be solved, is the positive definite symmetric coefficient matrix in the objective function, is the coefficient vector in the objective function, is the row full rank coefficient matrix in the equality constraint, is the coefficient vector in the equality constraint, is the coefficient matrix in the inequality constraints, is the coefficient vector in the inequality constraints.
[0089] S4: Based on the zeroing neurodynamics design rule, an exponential penalty strategy is introduced into the repetitive motion control scheme of the mobile robot based on the time-varying quadratic optimization problem to construct a continuous-time zeroing neurodynamics model:
[0090] S4.1: An exponential penalty strategy is introduced into the repetitive motion control scheme based on the time-varying quadratic optimization problem, and the inequality constraints in the repetitive motion control scheme are eliminated to obtain a repetitive motion control scheme with only equality constraints, which is expressed as follows:
[0091]
[0092] sG k+1 x k+1 =g k+1
[0093] in, is an exponential penalty function, and its expression is as follows:
[0094]
[0095] Among them, σ i (x k+1, t k+1 )=-H i,k+1 x k+1 +h i,k+1 represents the i-th auxiliary variable in the penalty function, H i,k+1 Represented by the matrix H k+1 The row vector composed of the i-th row elements of i,k+1 Represents vector h k+1 The i-th element of , μ and ν represent positive design parameters.
[0096] S4.2: According to the Lagrange multiplier method, the repetitive motion control scheme with only equality constraints is converted into a time-varying equation system problem, which is expressed as follows:
[0097] U k+1 y k+1+ω(x k+1 ,t k+1 )=0
[0098] Among them, U k+1 ,y k+1 and ω(x k+1 ,t k+1 ) are as follows:
[0099]
[0100]
[0101]
[0102] in, is the Lagrange multiplier vector.
[0103] S4.3: The time-varying equations are continuous-operated to obtain a continuous-time time-varying equations, the expression of which is as follows:
[0104] U(t)y(t)+ω(x(t),t)=0
[0105] S4.4: Define the error function of the continuous-time time-varying system of equations as ε(t)=U(t)y(t)+ω(x(t),t), and combine it with the zeroing neurodynamic design rule The continuous-time zeroing neural dynamics model is obtained, and its expression is as follows:
[0106]
[0107] Among them, η is a positive design parameter used to adjust the convergence rate; is an array of monotonically increasing odd excitation functions; M(x(t), t) represents the coefficient matrix, P(x(t), t) represents the time derivative matrix, u(x(t), t) represents the time derivative vector, and the expressions of M(x(t), t), P(x(t), t) and u(x(t), t) are as follows:
[0108]
[0109]
[0110]
[0111] S5: According to the five-step explicit linear multi-step rule and the continuous-time annihilation neural dynamics model, a discrete-time annihilation neural dynamics model is constructed, and the control variables of the mobile robot are calculated according to the discrete-time annihilation neural dynamics model.
[0112] In this embodiment, the five-step explicit linear multi-step rule is derived from the linear multi-step method and can be used for discrete-time solution of dynamic systems in the form of differential equations. The model based on the five-step explicit linear multi-step rule only needs to use the results and derivative values of the current and previous five discrete time points, and through an explicit linear combination form, it can predictively calculate the solution of the next discrete time point without using the information of the next discrete time point. The five-step explicit linear multi-step rule involves six adjacent discrete time points, that is, it contains five progressive steps, so it is called five steps. And because the five-step explicit linear multi-step rule does not need to use the information of the next discrete time point, it is called explicit.
[0113] In this embodiment, the expression of the five-step explicit linear multi-step rule is as follows:
[0114]
[0115] Combining the five-step explicit linear multi-step rule with the continuous-time annihilation neural dynamics model, a discrete-time annihilation neural dynamics model is obtained, which is expressed as follows:
[0116]
[0117] in, is the calculation assignment operator; is the truncation error of the discrete-time annihilation neural dynamics model, indicating that the discrete-time annihilation neural dynamics model has a sixth-order computational accuracy.
[0118] S6: Input the motion data of the mobile robot into the discrete time annihilation neural dynamics model to calculate the control variables of the mobile robot.
[0119] S7: Performing repeated motion control on the mobile robot according to the control variables.
[0120] In this embodiment, a lower computer controller is used to drive the mobile robot to perform repetitive motion control tasks according to the control variables.
[0121] In the specific implementation process, in order to ensure the convergence rate of the above discrete-time neural dynamics model, the power-S-shaped function is selected as the excitation function.
[0122] When conducting numerical experiments, the end effector of the mobile robot is expected to track a complex Lissajous graph path, which is expressed as follows:
[0123]
[0124] Among them, δ1, δ2 and δ3 are three constants that determine the expected path position.
[0125] Setting the time step Single repetition exercise duration t final =10s, the distance from the left and right driving wheels to the midpoint of the two wheel axles is a=0.32m, the distance from the midpoint of the two wheel axles to the connection point between the mobile platform and the robotic arm is b=0.1m, the radius of the left and right driving wheels is γ=0.1025m, and the other parameters are set to η=25, μ=0.004, ν=300, σ1=9, σ2=0.1, λ1=10, λ2=λ3=λ4=30.
[0126] In addition, the initial position p of the connection point between the mobile platform and the robotic arm connection (0) = [x connection (0);y connection The initial heading angle of the mobile platform is φ(0) = 0.1 rad, and the combined angle vector of the mobile robot is Θ(0) = [0, 0, π / 12, π / 3, π / 3, π / 3, π / 3, π / 3] T rad. The physical limit constraints of the mobile robot are set as follows:
[0127] Θ + =[∞,∞,0.524,2.356,1.088,6.283,6.283,6.283] T rad
[0128] Θ - =-[∞,∞,0.524,0.175,6.283,6.283,6.283,6.283] T rad
[0129]
[0130] Finally, the control variables are calculated by using the discrete-time zeroing neural dynamics model, thereby achieving high-precision real-time repetitive motion control of the mobile robot with physical limit constraints.
[0131] like Figure 4 As shown, it is the motion trajectory of the mobile robot of the present invention and the actual trajectory and expected path of the end effector. Figure 4 It can be seen that the actual trajectory of the end effector coincides with the expected Lissajous figure path, and the final state of the mobile robot coincides with the initial state after completing the path tracking task. Figure 5 is a diagram of the connection points between the mobile platform and the mechanical arm and the heading angle of the mobile platform of the present invention, Figure 6 is the element trajectory diagram of the combined angle vector of the mobile robot of the present invention, from Figure 5 and Figure 6 It can be seen that there are three important variables for achieving repetitive motion, namely the connection point p between the mobile platform and the robotic arm. connection,k+1 , the heading angle φ of the mobile platform k+1 and the combined angle vector Θ of the mobile robot k+1 , and finally return to their initial state, thus ensuring that after completing the closed path tracking task of the end effector, the purpose of coordinated repetitive motion control of the mobile robot can be achieved. In addition, Figure 7 is the element trajectory diagram of the combined velocity vector of the mobile robot of the present invention, Figure 8 is the element trajectory diagram of the tracking error vector of the end effector of the mobile robot of the present invention, from Figure 7 and Figure 8 It can be seen that the combined velocity vector of the mobile robot All element trajectories of converge to zero after the task is completed, and the end effector positioning error vector ∈ k+1 =r actual,k+1 -r desired,k+1 The X, Y, and Z axis components are all less than 7×10 -6 m, it is clear that the tracking accuracy of the end effector is high enough. In addition, from Figure 6 and Figure 7 It can be seen that all elements of the combined angle vector and the combined velocity vector of the mobile robot are limited to the given upper and lower limits of the physical constraints. In particular, the angle θ of the third joint of the mobile robot's manipulator is 3,k+1 The time-varying trajectory of reaches the upper limit of the physical limit many times, but never crosses the upper limit. Obviously, when the angle θ of the third joint 3,k+1 When the upper limit of the physical limit is about to be reached, the speed layer double-end inequality constraint obtained based on the exponential limit conversion strategy is successfully activated, which makes the speed of the third joint Slow down to avoid physical limits. It is worth noting that the average computation time of each update of the discrete time zeroing neural dynamics model is about 9.3×10 -4 s, which is much smaller than the time step used Thus, the requirements for strict real-time motion control of mobile robots can be met. In summary, these results well demonstrate the high-precision real-time repetitive motion control of mobile robots with physical limit constraints.
[0132] Example 3
[0133] See also Fig. 9 , this embodiment proposes a mobile robot repetitive motion control system based on discrete time neural dynamics, including:
[0134] The acquisition module is used to collect the motion data of the mobile robot.
[0135] The first building module is used to build the position layer kinematic equation, velocity layer kinematic equation and physical limit constraints of the mobile robot, and convert the physical limit constraints into velocity layer double-end inequality constraints.
[0136] In this embodiment, discrete time points are determined according to the update index and the time step, and the position layer kinematic equations are constructed at the discrete time points.
[0137] In this embodiment, the velocity layer kinematic equation is constructed based on the position layer kinematic equation.
[0138] In this embodiment, the physical limit constraint is constructed by discrete time-varying two-terminal inequality groups.
[0139] In this embodiment, based on the exponential limit conversion strategy, the physical limit constraint is converted into a velocity layer double-end inequality constraint.
[0140] The second building module is used to combine the position layer kinematic equation, the velocity layer kinematic equation and the velocity layer double-end inequality constraints to build a repetitive motion control scheme for the mobile robot based on a time-varying quadratic optimization problem.
[0141] In this embodiment, a repetitive motion control scheme of a mobile robot is designed by combining the kinematic equations of the position layer, the kinematic equations of the velocity layer, and the double-end inequality constraints of the velocity layer.
[0142] In this embodiment, the repetitive motion control scheme is converted into a time-varying quadratic optimization problem to obtain a repetitive motion control scheme based on the time-varying quadratic optimization problem.
[0143] The third building block is used to introduce an exponential penalty strategy into the repetitive motion control scheme of the mobile robot based on the time-varying quadratic optimization problem based on the zeroing neurodynamics design rule, and to construct a continuous-time zeroing neurodynamics model.
[0144] In this embodiment, an exponential penalty strategy is introduced into a repetitive motion control scheme based on a time-varying quadratic optimization problem, and the inequality constraints in the repetitive motion control scheme are eliminated to obtain a repetitive motion control scheme with only equality constraints.
[0145] According to the Lagrange multiplier method, the repetitive motion control scheme with only equality constraints is converted into a time-varying system of equations problem. The time-varying system of equations problem is subjected to a continuous operation to obtain a continuous-time system of equations problem, and an error function of the continuous-time system of equations problem is given. Combined with the zeroing neurodynamics design rule, a continuous-time zeroing neurodynamics model is obtained.
[0146] The fourth construction module is used to construct a discrete-time annihilation neural dynamics model according to the five-step explicit linear multi-step rule and the continuous-time annihilation neural dynamics model, and calculate the control variables of the mobile robot according to the discrete-time annihilation neural dynamics model.
[0147] The calculation module is used to calculate the control variables of the mobile robot according to the motion data of the mobile robot and the discrete time annihilation neural dynamics model.
[0148] The control module is used to perform repetitive motion control on the mobile robot according to the control variables.
[0149] In this embodiment, a lower computer controller is used to drive the mobile robot to perform repetitive motion control tasks according to the control variables.
[0150] By constructing the position layer kinematic equations, velocity layer kinematic equations and velocity layer double-end inequality constraints of the mobile robot, a repetitive motion control scheme for the mobile robot based on the time-varying quadratic optimization problem is designed, and on the basis of the repetitive motion control scheme of the mobile robot, an exponential penalty strategy and a five-step explicit linear multi-step rule are introduced to construct a discrete time zeroing neural dynamics model. In each calculation time interval, it can accurately predict the solution result at the next time point based on the current and previous known data information and through a single update calculation, thereby realizing high-precision real-time repetitive motion control of the mobile robot with physical limit constraints, effectively eliminating the non-repetitive motion phenomenon, and greatly improving the real-time motion, path tracking accuracy and physical limit constraint processing capability of the mobile robot.
[0151] The terms used in the drawings to describe positional relationships are only used for illustrative purposes and should not be construed as limiting this patent;
[0152] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.
Claims
1. A method for controlling repetitive motion of a mobile robot based on discrete-time neural dynamics, characterized in that: include: S1: Collect motion data of mobile robot; S2: Construct the position layer kinematic equation, velocity layer kinematic equation and physical limit constraints of the mobile robot, and convert the physical limit constraints into double-end inequality constraints of the velocity layer, including: S2.1: Determine the discrete time points of the mobile robot's motion and In the above example, the kinematic equation of the position layer is constructed, and its expression is as follows: in, To update the index, is the time step, is the final time point of the entire calculation time interval, is the nonlinear mapping function array, is the set of real numbers, is the combined angle vector of the mobile robot, is the actual position vector of the end effector, is the rotation angle set of the left and right driving wheels of the mobile platform, is the joint angle vector of the six-joint redundant manipulator fixed on the mobile platform; S2.2: According to the position layer kinematic equation, construct the velocity layer kinematic equation, the expression of which is as follows: in, is the heading angle of the mobile platform, is the combined velocity vector of the mobile robot, is the rotation speed set of the left and right driving wheels of the mobile platform, is the joint velocity vector of the six-joint redundant manipulator fixed on the mobile platform, is the matrix associated with the Jacobian is the vector containing the position error feedback of the end effector, is the time derivative of the desired path of the end effector, is the desired path of the end effector, is a configurable parameter; S2.3: The physical limit constraint is constructed by discrete time-varying two-terminal inequalities, which are expressed as follows: in, represents the lower limit of the combined angle vector of the mobile robot, represents the upper limit of the combined angle vector of the mobile robot, represents the lower limit of the combined velocity vector of the mobile robot, represents the upper limit of the combined velocity vector of the mobile robot; S2.4: Based on the exponential limit conversion strategy, the physical limit constraint is converted into a velocity layer double-end inequality constraint, the expression of which is as follows: in, represents the lower limit of the two-ended inequality constraint, represents the upper limit of a double-ended inequality constraint, No. Elements ),as well as No. Elements ), respectively, are defined as follows: in, represents the exponential function; It is a settable parameter used to adjust the deceleration range; parameter It is a configurable parameter used to adjust the critical area range that needs to be decelerated, that is, and ; S3: Combining the position layer kinematic equation, velocity layer kinematic equation and velocity layer double-end inequality constraints, a repetitive motion control scheme of a mobile robot based on a time-varying quadratic optimization problem is constructed, including: S3.1: Combining the position layer kinematic equation, velocity layer kinematic equation and velocity layer double-end inequality constraints, design a repetitive motion control scheme for the mobile robot, which is expressed as follows: min. s.t. in, The coefficient matrix representing the performance index of repeated motion, The coefficient vector representing the performance index of the repetitive motion, and The expressions are as follows: in, represents the auxiliary matrix related to the heading angle of the mobile platform, , and are all positive design parameters; is the connection point between the mobile platform and the robotic arm. represents the coordinate of the connection point on the X axis, Represents the coordinate of the connection point on the Y axis; represents the initial state of the connection point between the mobile platform and the robotic arm, is the initial state of the heading angle of the mobile platform, represents the initial state of the joint angle vector of the robot arm, is the distance from the left driving wheel and the right driving wheel to the midpoint of the two wheel axles; From the midpoint of the two axles to the connection point The distance is the radius of the left and right driving wheels; , represents the identity matrix; S3.2: Convert the repetitive motion control scheme into a time-varying quadratic optimization problem, and obtain a repetitive motion control scheme based on the time-varying quadratic optimization problem, the expression of which is as follows: min. s.t. in, represents the transpose operator of a vector or matrix, is the decision variable vector to be solved, is the positive definite symmetric coefficient matrix in the objective function, is the coefficient vector in the objective function, is the row full rank coefficient matrix in the equality constraint, is the coefficient vector in the equality constraint, is the coefficient matrix in the inequality constraints, is the coefficient vector in the inequality constraint; S4: Based on the zeroing neurodynamics design rule, an exponential penalty strategy is introduced into the repetitive motion control scheme of the mobile robot based on the time-varying quadratic optimization problem to construct a continuous-time zeroing neurodynamics model; S5: constructing a discrete-time annulization neural dynamics model based on the five-step explicit linear multi-step rule and the continuous-time annulization neural dynamics model; S6: inputting the motion data of the mobile robot into the discrete time annulization neural dynamics model to calculate the control variables of the mobile robot; S7: Performing repeated motion control on the mobile robot according to the control variables.
2. The method for repetitive motion control of a mobile robot based on discrete-time neural dynamics according to claim 1, characterized in that: The specific steps of S4 include: S4.1: An exponential penalty strategy is introduced into the repetitive motion control scheme based on the time-varying quadratic optimization problem, and the inequality constraints in the repetitive motion control scheme are eliminated to obtain a repetitive motion control scheme with only equality constraints, which is expressed as follows: min. s.t. in, is an exponential penalty function, and its expression is as follows: in, represents the penalty function i Auxiliary variables, Represented by the matrix No. The row vector composed of row elements, Representation vector No. elements, and represents a positive design parameter; S4.2: According to the Lagrange multiplier method, the repetitive motion control scheme with only equality constraints is converted into a time-varying equation system problem, which is expressed as follows: in, , and The expressions are as follows: in, is the Lagrange multiplier vector; S4.3: The time-varying equations are continuous-operated to obtain a continuous-time time-varying equations, the expression of which is as follows: S4.4: The error function of the continuous-time time-varying equations problem is defined as , and combined with the zeroing neurodynamic design principle , we get the continuous-time zeroing neural dynamics model, which is expressed as follows: in, is a positive design parameter used to adjust the convergence rate; is an array of monotonically increasing odd excitation functions; represents the coefficient matrix, represents the time derivative matrix, represents the time derivative vector, and , and The expressions are as follows: 。 3. The method for controlling repetitive motion of a mobile robot based on discrete-time neural dynamics according to claim 2, characterized in that: In S5, a discrete-time annihilation neural dynamics model is constructed according to the five-step explicit linear multi-step rule and the continuous-time annihilation neural dynamics model, and its expression is as follows: in, is the calculation assignment operator; ), ; is the truncation error of the discrete-time annihilation neural dynamics model, indicating that the discrete-time annihilation neural dynamics model has a sixth-order computational accuracy.
4. The method for controlling repetitive motion of a mobile robot based on discrete-time neural dynamics according to any one of claims 1 to 3, characterized in that: In S7, the lower computer controller is used to drive the mobile robot to perform the repetitive motion control task according to the control variable.
5. A mobile robot repetitive motion control system based on discrete-time neural dynamics, applied to a mobile robot repetitive motion control method based on discrete-time neural dynamics as claimed in any one of claims 1 to 4, characterized in that: include: A collection module, used to collect motion data of the mobile robot; The first building module is used to build the position layer kinematic equation, velocity layer kinematic equation and physical limit constraints of the mobile robot, and convert the physical limit constraints into velocity layer double-end inequality constraints; The second building module is used to construct a repetitive motion control scheme of the mobile robot based on a time-varying quadratic optimization problem by combining the position layer kinematic equation, the velocity layer kinematic equation and the velocity layer double-end inequality constraint; The third building block is used to introduce an exponential penalty strategy into the repetitive motion control scheme of the mobile robot based on the time-varying quadratic optimization problem based on the zeroing neural dynamics design rule, and to build a continuous-time zeroing neural dynamics model; A fourth construction module is used to construct a discrete-time annulization neural dynamics model according to the five-step explicit linear multi-step rule and the continuous-time annulization neural dynamics model; A calculation module, used for calculating the control variables of the mobile robot according to the motion data of the mobile robot and the discrete time annulization neural dynamics model; The control module is used to perform repetitive motion control on the mobile robot according to the control variables.
6. An electronic device, characterized in that: It comprises a processor and a memory; the memory is used to store a program; the processor executes the program to implement the mobile robot repetitive motion control method based on discrete-time neural dynamics as described in any one of claims 1 to 4.
7. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement the mobile robot repetitive motion control method based on discrete-time neural dynamics as described in any one of claims 1 to 4.
