Preset time formation control method for robot model with output error constraints

By constructing the Lyapunov function and preset time filter, combined with the radial basis neural network, the problem of insufficient response speed and accuracy in the formation control of the robot model is solved, and the system stability and error convergence effect within the preset time is achieved .

CN118348806BActive Publication Date: 2024-11-15YANGU (HANGZHOU) E-COMMERCE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410608790.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-16
Publication Date
2024-11-15
Estimated Expiration
2044-05-16

Smart Images

  • Figure CN118348806B_ABST
    Figure CN118348806B_ABST
Patent Text Reader

Abstract

The present invention discloses a robot model preset time formation control method with output error constraints, wherein the establishment process of the robot model preset time formation controller includes establishing a robot control model, establishing a Lyapunov function, establishing a preset time filter, establishing an adaptive update law, and establishing a preset time controller. The existing control method for robot model formation control does not consider the response speed problem of the robot model formation during the control process, nor can it guarantee that the system output state tracking error will converge to a desired accuracy range. The present invention designs a preset time filter and a preset time controller, and constrains the system output error, so that the control system can be stable within a preset time, and ensures that the output state tracking error can converge to a desired control accuracy range. It provides strong technical support for the high response speed and high precision control of robot model formation control.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of robot control, in particular to a robot model preset time formation control method with output error constraints. Background Art

[0002] Robot formation control refers to a technology that controls multiple robots through cooperation so that they form an orderly formation and complete specific tasks. Robot formation control technology is increasingly used in many fields such as industrial automation, logistics and transportation, and rescue missions. In these applications, robot formations can improve work efficiency, reduce costs, and enhance safety. In recent years, with the vigorous development of technologies such as sensors, artificial intelligence, and intelligent control, omnidirectional mobile robots have gradually become one of the important development directions in the future field of robotics due to their multi-directional mobility, high maneuverability, good interactivity, and intelligence.

[0003] Robot model formation control aims to develop an adaptive fuzzy control method to achieve stable motion of robot formations and enable them to adapt to changes and uncertainties in the environment. Specifically, the control goal of robot model formation control is to design an adaptive controller to ensure that all signals in the closed-loop system are in a semi-global eventually consistent and bounded state, and the output of the system can effectively track the reference signal. At present, there are many control algorithms and techniques for robot model formation control. The mainstream control methods include radial basis neural networks, fuzzy logic systems, and PID controllers. However, the existing technology still has the following problems:

[0004] First, in the existing control methods for robot model formation control, only a controller can be designed to make the system output effectively track the reference signal. The existing technology does not consider the response speed of the robot model formation during the control process. The response speed is a key indicator for evaluating the dynamic characteristics of the controlled system. In actual engineering applications, people usually hope that the state of the controlled system can quickly converge to a stable state.

[0005] Second, in the existing control methods for robot model formation control, although the control system can converge the system output state tracking error to a certain range, it cannot guarantee that the system output state tracking error will converge to the desired accuracy range. In practical applications, people pursue not only high response speed of the system control process, but also high control accuracy. Summary of the invention

[0006] In order to address the deficiencies in the prior art, the present invention designs a robot model preset time formation control method with output error constraints, so that the control system can be stable within a preset time and ensure that the output state tracking error converges to a desired control accuracy range.

[0007] The present invention provides a robot model preset time formation control method with output error constraints, which is controlled by using a robot model preset time formation controller. The establishment process of the robot model preset time formation controller includes establishing a robot control model, establishing a Lyapunov function, establishing a preset time filter, establishing an adaptive update law, and establishing a preset time formation controller. First, a robot model is constructed, and on this basis, a Lyapunov function and a preset time filter are constructed, and finally an adaptive update law and a preset time controller of the control system are obtained, which are specifically as follows:

[0008] Step 1, establishing a dynamic model of the omnidirectional mobile robot;

[0009] Step 2, calculating the output state tracking error of the robot according to the dynamic model;

[0010] Step 3, using radial basis neural network to approximate the unknown nonlinear functions contained in the dynamic model, so as to establish the Lyapunov function; the barrier Lyapunov function is introduced into the Lyapunov function to deal with the output state tracking error constraint problem;

[0011] Step 4: construct a preset time filter, design an adaptive law and a preset time formation controller according to the Lyapunov function.

[0012] Furthermore, in step 1, a dynamic model of the omnidirectional mobile robot is established, wherein the dynamic model of the i-th omnidirectional mobile robot is described by the following differential equations:

[0013]

[0014] and They represent the position and velocity states of the i-th omnidirectional mobile robot respectively;

[0015] represents the robot coordinate system X of the i-th omnidirectional mobile robot r The positive direction of the axis is the same as the world coordinate system X w The angle between the positive directions of the axes;

[0016] y i represents the output state of the i-th omnidirectional mobile robot;

[0017] u i represents the preset time formation controller of the i-th omnidirectional mobile robot;

[0018] f i,1 、f i,2 、f i , g i They are and abbreviation of;

[0019] f i,1 and f i,2 Represents the problems encountered by omnidirectional mobile robots in actual motion, namely the possible generation of unknown nonlinear dynamics due to the difference between motion conditions and modeling assumptions;

[0020] f i and g i represent the nonlinear dynamics and control input gain matrices inherent to the omnidirectional mobile robot, respectively.

[0021] Furthermore, in step 3, a radial basis neural network is used to approximate the unknown nonlinear functions contained in the dynamic model, specifically:

[0022] The inherent nonlinear dynamics of omnidirectional mobile robots i It is expressed as follows:

[0023]

[0024] Represents the transformation matrix between two coordinate systems;

[0025] represents the static friction force on the Mecanum wheel;

[0026] D θ represents the viscous friction coefficient of the Mecanum wheel;

[0027] J + and J represent the Jacobian matrices of the forward kinematics model and inverse kinematics model of the omnidirectional mobile robot, respectively;

[0028] Represents the angular velocity of each wheel of the omnidirectional mobile robot;

[0029] represents the angular acceleration of the wheat wheel, and M is its gain matrix.

[0030] In order to achieve the fitting of unknown nonlinear dynamics in the system, the following RBF-NN is designed:

[0031]

[0032] represents the ideal neural network weight matrix;

[0033] S i,t (Z i,t ) is the output vector of the neural network,

[0034] εi,t (Z i,t ) is the neural network fitting error vector, which satisfies is a positive constant;

[0035] Γ i,t (Z i,t ) is the unknown nonlinear dynamics in the system;

[0036] Z i,t is the input vector of the neural network,

[0037] Furthermore, in step 3, the Lyapunov function is established based on this. Specifically, for an omnidirectional mobile robot system whose dynamic model is a second-order model, the Lyapunov function needs to be constructed twice;

[0038] The first construction of the Lyapunov function is expressed in the following form:

[0039]

[0040] V i,b is the barrier Lyapunov function, which is used to deal with the output error constraint problem. The barrier Lyapunov function V i,b It is expressed by the following formula:

[0041]

[0042] N represents the number of agents;

[0043] k i,b is a constraint on the output error and is a constant vector;

[0044] z i,1 is the output state tracking error;

[0045] γ i,1 Design parameters for the adaptive update law;

[0046] is the ideal adjustment scalar The estimated error with respect to its estimated value;

[0047] The second construction of the Lyapunov function can be expressed as follows:

[0048]

[0049] z i,2 represents the virtual tracking error;

[0050] s i,2 represents the filtering error;

[0051] γi,2 Design parameters for the adaptive update law;

[0052] is the ideal adjustment scalar The estimated error from its estimated value.

[0053] Furthermore, in step 4, a preset time filter is constructed, which is expressed by the following differential equation:

[0054]

[0055] h∈(0,1), is a design parameter;

[0056] δ is the filter design parameter;

[0057] T d It is a predetermined time;

[0058] α i,1 It is a virtual controller;

[0059] Design of adaptive law and preset time formation controller

[0060] The first formulation of the adaptive law is expressed in the following form:

[0061]

[0062] The second construction of the adaptive law is expressed in the following form:

[0063]

[0064] Where S i,τ is the output vector of the neural network;

[0065] The final preset time formation controller is obtained, which is expressed by the following formula:

[0066]

[0067] Compared with the existing working technology, the present invention has the following beneficial effects:

[0068] First, in the existing control method of robot model formation control, only a controller that enables the system output to effectively track the reference signal can be designed, and the response speed of the robot model formation in the control process is not considered. However, the present invention considers that the response speed is a key indicator for evaluating the dynamic characteristics of the controlled system, so a preset time filter and a preset time controller are designed to enable the control system to be stable within a preset time, and the state of the system can quickly converge to a stable state, and the convergence time can be preset.

[0069] Second, the existing control methods for robot model formation control cannot guarantee that the system output state tracking error will converge to the desired accuracy range. However, the present invention constrains the system output error, which can ensure that the output state tracking error can converge to a desired control accuracy range. This provides a strong guarantee for the high-precision control of the robot model formation control. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 This is a simplified diagram of the kinematic analysis of an omnidirectional mobile robot.

[0071] Figure 2 This is a schematic diagram of the communication topology between robots.

[0072] Figure 3 It is the motion trajectory diagram of the multi-agent omnidirectional mobile robot system formation in the XY plane.

[0073] Figure 4 It is the position state tracking error diagram of each robot in the X direction.

[0074] Figure 5 It is the velocity state tracking error diagram of each robot in the X direction.

[0075] Figure 6 It is the position state tracking error diagram of each robot in the Y direction.

[0076] Figure 7 It is the velocity state tracking error diagram of each robot in the Y direction.

[0077] Figure 8 It is the output state tracking error diagram of each robot in the X direction.

[0078] Fig. 9 It is the output state tracking error diagram of each robot in the Y direction. DETAILED DESCRIPTION

[0079] The present invention combines adaptive backstepping and dynamic surface control technology to propose an output error constraint control strategy with preset time stability based on output feedback. Under this control strategy, the preset time stability of the system is achieved and the output error constraint is satisfied. In this embodiment, a multi-agent system consisting of four robots is selected as the control object. The robots are omnidirectional mobile robots. The schematic diagram of the communication topology structure between the omnidirectional mobile robots is shown in the figure. Figure 2As shown, L1 represents the leader, and F1, F2, F3, and F4 represent four followers. The dynamic model of the robot is established. The output state tracking error of the robot can be obtained by analyzing the dynamic model of the robot. The radial basis neural network is used to approximate the unknown nonlinear function, so as to establish the Lyapunov function. The obstacle Lyapunov function is introduced to deal with the output state tracking error constraint problem, and then the preset time filter is constructed. Based on the above work, the adaptive law and the preset time formation controller can be obtained;

[0080] The present invention provides a robot model preset time formation method with output error constraints, and the specific steps are as follows:

[0081] The process of establishing the preset time formation controller is as follows:

[0082] A. Establishment of control model of omnidirectional mobile robot

[0083] In a multi-agent omnidirectional mobile robot system, in order to distinguish multiple omnidirectional mobile robots, each robot is numbered, and the dynamic model of the i-th omnidirectional mobile robot is given. The schematic diagram of the dynamic analysis of the omnidirectional mobile robot is shown in Figure 1 As shown, the robot dynamics model is described by the following differential equations:

[0084]

[0085] and They represent the position and velocity states of the i-th omnidirectional mobile robot respectively;

[0086] represents the robot coordinate system X of the i-th omnidirectional mobile robot r The positive direction of the axis is the same as the world coordinate system X w The angle between the positive directions of the axes;

[0087] y i represents the output state of the i-th omnidirectional mobile robot;

[0088] u i represents the preset time formation controller of the i-th omnidirectional mobile robot;

[0089] f i,1 、f i,2 、f i , g i They are and Abbreviation of f i,1 and f i,2 Represents the problems encountered by omnidirectional mobile robots in actual motion, namely the possible generation of unknown nonlinear dynamics due to the difference between motion conditions and modeling assumptions;

[0090] f i and g i They represent the inherent nonlinear dynamics and control input gain matrices of the omnidirectional mobile robot. These characteristics are derived from its unique structural features and can be expressed by the following mathematical formula:

[0091]

[0092] Represents the transformation matrix between two coordinate systems;

[0093] represents the static friction force on the Mecanum wheel;

[0094] D θ represents the viscous friction coefficient of the Mecanum wheel;

[0095] J + and J represent the Jacobian matrices of the forward kinematics model and inverse kinematics model of the omnidirectional mobile robot, respectively;

[0096] Represents the angular velocity of each wheel of the omnidirectional mobile robot;

[0097] represents the angular acceleration of the wheat wheel, and M is its gain matrix.

[0098] B. Establishment of Lyapunov function

[0099] For an omnidirectional mobile robot system whose dynamic model is a second-order model, we only need to construct the Lyapunov function twice to obtain its controller and adaptive law.

[0100] The first construction of the Lyapunov function can be expressed in the following form:

[0101]

[0102] V i,b is the barrier Lyapunov function, which is used to deal with the output error constraint problem and can be expressed by the following formula:

[0103]

[0104] N represents the number of agents;

[0105] k i,b is a constraint on the output error and is a constant vector;

[0106] z i,1 is the output state tracking error;

[0107] γ i,1Design parameters for the adaptive update law;

[0108] is the ideal adjustment scalar The estimated error from its estimated value.

[0109] The second construction of the Lyapunov function can be expressed as follows:

[0110]

[0111] z i,2 represents the virtual tracking error;

[0112] s i,2 represents the filtering error;

[0113] γ i,2 Design parameters for the adaptive update law;

[0114] is the ideal adjustment scalar The estimated error from its estimated value.

[0115] C. Preset time filter establishment

[0116] The preset time filter is expressed by the following differential equation:

[0117]

[0118] h∈(0,1), is a design parameter;

[0119] δ is the filter design parameter;

[0120] T d It is a predetermined time;

[0121] α i,1 It is a virtual controller.

[0122] D. Adaptive Law and Controller Establishment

[0123] The first construction of the adaptive law can be expressed in the following form:

[0124]

[0125] The second construction of the adaptive law can be expressed in the following form:

[0126]

[0127] Where S i,τ is the output vector of the neural network.

[0128] Based on the previous work, we can get the final preset time formation controller, which can be expressed by the following formula:

[0129]

[0130] The simulation results are as follows Figure 3-9 shown. Figure 3 The figure shows the formation motion effect of a multi-agent omnidirectional mobile robot system. The black straight line represents the motion trajectory generated by the reference signal, and the other lines represent the motion trajectory of each agent. It can be seen from the figure that the four OMRs can track the reference signal with a certain position offset. Figure 4 The position status tracking effect of each robot in the X direction is shown. Figure 5 The speed state tracking effect of each robot in the X direction is shown. Figure 6 The position status tracking effect of each robot in the Y direction is shown. Figure 7 The speed state tracking effect of each robot in the Y direction is shown. From the above four figures, it can be seen that the robot formation system formed a formation at about 9s and has maintained the formation movement since then. Specifically, the position states of OMR in the X and Y directions maintain the preset offset with the corresponding components of the reference signal. Figure 8 The output state tracking effect of each robot in the X direction is shown. Fig. 9 The output state tracking effect of each robot in the Y direction is shown. From these two figures, it can be seen that the output state tracking error of each OMR converges to a certain limit in about 9s and meets the constraint conditions. From the above simulation result figures, it can be seen that the tracking errors of the position state and speed state of each robot in the omnidirectional mobile robot formation control system in the X and Y directions are stable within the preset time, achieving the expected preset time formation control.

[0131] The present invention is not limited to this embodiment, and any equivalent concepts or changes within the technical scope disclosed by the present invention are included in the protection scope of the present invention.

Claims

1. Preset time formation control method for robot model with output error constraints, It is characterized in that The steps include: Step 1: Establish a dynamic model of the omnidirectional mobile robot, where the dynamic model of the i-th omnidirectional mobile robot is described by the following differential equations: and They represent the position and velocity states of the i-th omnidirectional mobile robot respectively; represents the robot coordinate system X of the i-th omnidirectional mobile robot r The positive direction of the axis is the same as the world coordinate system X w The angle between the positive directions of the axes; y i represents the output state of the i-th omnidirectional mobile robot; u i represents the preset time formation controller of the i-th omnidirectional mobile robot; f i,1 、f i,2 、f i , g i They are and abbreviation of; f i,1 and f i,2 Represents the problems encountered by omnidirectional mobile robots in actual motion, namely the possible generation of unknown nonlinear dynamics due to the difference between motion conditions and modeling assumptions; f i and g i denote the inherent nonlinear dynamics and control input gain matrices of the omnidirectional mobile robot respectively; Step 2, calculating the output state tracking error of the robot according to the dynamic model; Step 3, using a radial basis neural network to approximate the unknown nonlinear function contained in the dynamic model, so as to establish a Lyapunov function; specifically, the dynamic model is an omnidirectional mobile robot system of a second-order model, and the Lyapunov function needs to be constructed twice; The first construction of the Lyapunov function is expressed in the following form: V i,b is the barrier Lyapunov function, which is used to deal with the output error constraint problem. The barrier Lyapunov function V i,b It is expressed by the following formula: N represents the number of agents; k i,b is a constraint on the output error and is a constant vector; z i,1 is the output state tracking error; g i,1 Design parameters for the adaptive update law; is the ideal adjustment scalar The estimated error with respect to its estimated value; The second construction of the Lyapunov function is expressed in the following form: z i,2 represents the virtual tracking error; s i,2 represents the filtering error; γ i,2 Design parameters for the adaptive update law; is the ideal adjustment scalar The estimated error with respect to its estimated value; Step 4: construct a preset time filter, and design an adaptive law and a preset time formation controller according to the Lyapunov function.

2. The robot model preset time formation control method with output error constraint according to claim 1, It is characterized in that In step 3, the radial basis neural network is used to approximate the unknown nonlinear function contained in the dynamic model, specifically: The inherent nonlinear dynamics of omnidirectional mobile robots i It is expressed as follows: Represents the transformation matrix between two coordinate systems; represents the static friction force on the Mecanum wheel; D θ represents the viscous friction coefficient of the Mecanum wheel; J + and J represent the Jacobian matrices of the forward kinematics model and inverse kinematics model of the omnidirectional mobile robot, respectively; Represents the angular velocity of each wheel of the omnidirectional mobile robot; represents the angular acceleration of the wheat wheel, and M is its gain matrix; In order to achieve the fitting of unknown nonlinear dynamics in the system, the following RBF-NN is designed: represents the ideal neural network weight matrix; S i,t (Z i,t ) is the output vector of the neural network, ε i,t (Z i,t ) is the neural network fitting error vector, which satisfies is a positive constant; Γ i,t (Z i,t ) is the unknown nonlinear dynamics in the system; Z i,t is the input vector of the neural network, 3. The robot model preset time formation control method with output error constraint according to claim 2, It is characterized in that The preset time filter constructed in step 4 is expressed by the following differential equation: h∈(0,1), is a design parameter; δ is the filter design parameter; T d It is a predetermined time; α i,1 It is a virtual controller; Design of adaptive law and preset time formation controller The first formulation of the adaptive law is expressed in the following form: The second construction of the adaptive law is expressed in the following form: Where S i,τ is the output vector of the neural network; The final preset time formation controller is obtained, which is expressed by the following formula: