Slip steering mobile robot model prediction control method based on adaptive sampling time
By adopting the model prediction control method of adaptive sampling time in slip-steering mobile robots, the sampling time is dynamically adjusted to respond to angular velocity changes, solving the computational burden and tracking performance problems caused by fixed sampling time in traditional MPC algorithms, and achieving more Efficient path tracking and better adaptability.
Patent Information
- Application Number
- CN202410940359.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-11-05
AI Technical Summary
Traditional Model Predictive Control (MPC) algorithms use fixed sampling time in slip-steering mobile robot path tracking, resulting in a high computational burden under short sampling intervals, but tracking performance may be affected as the sampling interval increases, and for different paths Limited adaptability.
The sliding steering mobile robot model prediction control method with adaptive sampling time is adopted. By matching the lidar scanning and speedometer computer robot posture, a kinematic model is established, and the cost function is minimized within the constraints using the quadratic planning method. Dynamically adjust the sampling time to respond to angular velocity changes to ensure tracking performance and control stability.
It effectively balances the tracking performance and control stability, improves the path tracking capability of mobile robots in complex trajectories and different terrains, reduces the threshold range of sampling time, and improves computing efficiency and tracking accuracy.
Smart Images

Figure CN118897464A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of autonomous navigation of mobile robots, and in particular relates to a model predictive control method, medium and equipment for a skid-steering mobile robot based on adaptive sampling time. Background Art
[0002] Path tracking is a fundamental challenge in motion control of a skid-steer mobile robot and a key element of unmanned navigation technology, the main goal of which is to enable a skid-steer mobile robot to accurately follow a specified path. However, due to the existence of non-integrable (nonholonomic) constraints in wheeled mobile robots, these constraints must be carefully considered in path planning and tracking.
[0003] Model predictive control (MPC) technology has the ability to be dynamic, manage multivariable problems, constraints, and optimize cost functions. As an efficient control method, it is widely used in the study of longitudinal and lateral vehicle control. However, in the study of the influence of control variables on path tracking performance in traditional MPC algorithms, fixed sampling time is mainly used for path tracking, and the selection of sampling time is usually based on empirical knowledge. The fixed sampling time in MPC can improve the tracking performance at shorter sampling intervals, but it will bring a huge computational burden. On the contrary, as the sampling interval increases, the operation speed will increase, but once interference occurs, the tracking performance may be affected. The disadvantage of this method is that it has limited adaptability to different paths. Summary of the invention
[0004] The present invention focuses on the problems existing in the above-mentioned prior art and provides a model predictive control method, medium and device for a skid-steering mobile robot based on adaptive sampling time.
[0005] A first aspect of the present invention provides a model predictive control method for a skid-steering mobile robot based on adaptive sampling time, comprising:
[0006] Step S100: Initialize sampling time T;
[0007] Step S200: By matching the current laser radar scan with the known map, the robot's position is calculated with the wheel speed meter, and the state variable is defined as P = (x, y, Ψ) T , control variable u=(v,ω), where the control variable is the velocity vector at the center of mass of the robot, v is the velocity, and ω is the angular velocity. The kinematic model of the wheeled robot is established, which is specifically expressed as follows:
[0008] in is the velocity of the robot on the x-axis, is the component velocity of the robot on the y-axis, is the robot angular velocity;
[0009] Step S300: discretize the robot kinematic model, set the predictive control model parameters including the prediction time domain, control time domain, initial sampling time, input weight matrix, state error weight matrix and constraint conditions, and construct a cost function including state error and control input;
[0010] Step S400: Use the quadratic programming method to solve the predictive control model, minimize the cost function within the constraints to obtain the optimal control input sequence, and predict the state vector to be X(k+1)=A·X(k)+B·u(k), where X(k) is the state vector and u(k) is the control input vector. Update the state equation X(k)=X(k+1), where B is the input matrix and A is the state transfer matrix.
[0011] Step S500: Apply the first control input in the optimal control input sequence to the robot controller, store the predicted state as the past state, and store the control input as the past input, and use the calculated angular velocity change rate in the optimal control input to calculate the sampling time T to be used when the vehicle moves next time;
[0012] Step S600: Set T=T(k+1) to determine whether the calculated sampling time is within a preset threshold. If so, set it as a new sampling time. If it exceeds the maximum value of the threshold or is less than the minimum value of the threshold, set the sampling time to the maximum value or minimum value of the threshold accordingly.
[0013] Step S700: Go to step S200 until the control ends.
[0014] Preferably, the step S200 further comprises:
[0015] The forward kinematics relationship of the wheeled robot is established, which is specifically expressed in the form of the following matrix product:
[0016] Among them, V l and V r is the left and right wheel speed.
[0017] Preferably, the step S300 further comprises:
[0018] The existing predictive control model, the current state variables of the MPC control system and the future control variables are used to predict the future output of the system and compare it with the expected system output to obtain the cost function:
[0019] J=[F out (M,C state ,F control )-D out ] 2
[0020] Where J is the cost function, F out is the future output, M is the model, C state is the current state, F control is the future control variable, D out is the expected output.
[0021] Preferably, the step S400 further comprises:
[0022] Add upper and lower velocity limits, and an upper limit on the angular velocity as system constraints for the quadratic programming solver.
[0023] Preferably, step S500 further includes:
[0024] The next sampling time T is calculated according to the angular velocity change rate. Specifically, as the angular velocity change rate increases, the corresponding sampling time will be shorter, and as the angular velocity change rate decreases, the corresponding sampling time will be longer.
[0025] Preferably, step S500 further includes:
[0026] The specific formula for calculating the next sampling time is: T(k+1)=T(k)+sign(ξ-C)·S, where:
[0027]
[0028] α, S and C are constant parameters, which are set based on experience. α is the sensitivity of the rate of change of the angular velocity to the adjustment of the sampling time. S is the step size of the increase or decrease of the sampling time. Sign (ξ-C) is used to determine whether the sampling time should be increased or decreased in the next step. The sign function value 1, 0, -1 corresponds to an increase, no change and a decrease in the number of sampling times in the next step, respectively. is the rate of change of angular velocity in the control variable of the predictive control model.
[0029] Preferably, the skid steer mobile robot has four wheels, which are driven by four independent motors respectively, and the motion state of the robot is controlled by controlling the speed.
[0030] A second aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the aforementioned methods.
[0031] According to a third aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of any of the aforementioned methods when executed by the processor.
[0032] Compared with the prior art, the present invention has the following beneficial effects:
[0033] 1) Adaptive sampling time is combined with MPC for the first time to balance the relationship between tracking performance and control stability. It solves the problem that fixed sampling time in traditional MPC has high tracking performance at short sampling intervals, but a large computational burden. On the contrary, as the sampling interval increases, the calculation speed will increase, but once interference occurs, the tracking performance may be affected.
[0034] 3) Adjust the sampling time according to the rate of change of the optimal control quantity in the MPC algorithm and the real-time status of the path to ensure that the sampling time meets the different requirements of the system dynamics and effectively responds to the period of drastic changes and weakening of the angular velocity, so that the mobile robot has the ability to dynamically respond to various terrains and challenge complex trajectories when facing curved paths;
[0035] 2) The proposed model predictive control path tracking algorithm with adaptive sampling time (AST-MPC) can dynamically adjust the sampling time in real time according to the optimal control inputs calculated within the MPC framework and evaluate the rate of change of these control inputs. This approach helps to effectively control the robot's path tracking and gradually converge the sampling time to the minimum or maximum value. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 Flowchart of model predictive control method for skid-steering mobile robot based on adaptive sampling time;
[0037] Figure 2 Schematic diagram of the simplified kinematic model of a wheeled robot;
[0038] Figure 3 Schematic diagram of MPC structure;
[0039] Figure 4 Schematic diagram of the path of the skid steer robot in scenario 1 and scenario 2;
[0040] Figure 5 Comparison of tracking performance of the AST-MPC algorithm of the present invention and the traditional MPC algorithm with fixed sampling time in scenario 1 and scenario 2;
[0041] Figure 6 The appearance of the skid steer mobile robot hardware. DETAILED DESCRIPTION
[0042] The technology described below can be transformed in many ways and can have many embodiments, which are described in detail with reference to the accompanying drawings in specific embodiments. However, this does not mean that the technology described below is limited to specific embodiments. It should be understood that the present invention includes all similar modifications, equivalents and substitutions without departing from the spirit and technical scope of the technology described below.
[0043] like Figure 1 As shown, the present invention provides a model predictive control method for a skid-steering mobile robot based on adaptive sampling time, comprising:
[0044] Step S100: Initialize sampling time T;
[0045] Step S200: By matching the current laser radar scan with the known map and using the wheel speed meter to calculate the robot's position and posture, a kinematic model of the wheeled robot is established;
[0046] Step S300: discretize the robot kinematic model, set the predictive control model parameters including the prediction time domain, control time domain, initial sampling time, input weight matrix, state error weight matrix and constraint conditions, and construct a cost function including state error and control input;
[0047] Step S400: using a quadratic programming method to solve the predictive control model, minimizing the cost function within the constraints to obtain the optimal control input sequence;
[0048] Step S500: Apply the first control input in the optimal control input sequence to the robot controller, store the predicted state as the past state, and store the control input as the past input, and use the calculated angular velocity change rate in the optimal control input to calculate the sampling time T to be used when the vehicle moves next time;
[0049] Step S600: Set T=T(k+1) to determine whether the calculated sampling time is within a preset threshold. If so, set it as a new sampling time. If it exceeds the maximum value of the threshold or is less than the minimum value of the threshold, set the sampling time to the maximum value or minimum value of the threshold accordingly.
[0050] Step S700: Go to step S200 until the control ends.
[0051] The skid-steer mobile robot has four wheels, which are driven by four independent motors respectively, and the motion state of the robot is controlled by controlling the speed.
[0052] More specifically, the SSMR kinematic model construction process is as follows:
[0053] The simplified kinematic model of SSMR is as follows: Figure 2 As shown, the relationship between chassis angular velocity, speed and circumference radius is shown. Define two coordinate systems, xo′y is the robot coordinate system, and XOY is the Cartesian coordinate system. The four wheels i (i=1,2,3,4) are driven by four independent motors M i(i=1,2,3,4) drive, and control the robot's motion state by controlling the speed. Among them, COG is the geometric center, COM is the center of mass, and ICR is the rotation center that changes with time and is located on the line of COG along the y-axis. The speed of the wheel v i is the lateral sliding velocity v iy and the preset target speed v ix The (v, omega) at COM is the velocity vector of SSMR, υ is the linear velocity perpendicular to the ICR-COM line segment, and ω is the angular velocity. The wheelbase of the left and right wheels is c, and the distances from point COM to the rear and front ends of the chassis are a and b respectively. The matrix product form of the SSMR forward kinematics relationship is shown in the following formula:
[0054]
[0055] Among them, V l and V r is the left and right wheel speed. The state variables defined are P = (x, y, Ψ) T , the control variable is u=(v,ω). The kinematic model description of SSMR is shown in the following formula:
[0056]
[0057] Preferably, the MPC algorithm principle involved in step S400 is:
[0058] The principle of MPC algorithm is as follows Figure 3 The MPC structure is shown in the schematic diagram, which consists of three parts: prediction model, rolling optimization and feedback correction. The application of MPC in path planning requires the definition of an expected reference path with the minimum tracking error for the SSMR, as well as an operating path consisting of past outputs and predicted future outputs. Figure 2 In the example, k is the current time, and MPC collects data from measurements and prediction models to predict [k+1,…,k+N p ] in order to obtain the predicted path. Then, solve the objective function to obtain [k+1,…,k+N p ] is the control variable sequence. However, the optimal control of the system can be achieved by only passing the control variables predicted in k+1.
[0059] The prediction model is difficult to accurately describe the system, and there may be disturbances, so rolling optimization is used. The core idea of rolling optimization is to obtain the optimal control strategy at each sampling moment by solving the local optimal control problem based on the current system state and measurement information. Then, at the next sampling moment, the optimization process is repeated.
[0060] The existing predictive control model, the current state variables of the system and the future control variables are used to predict the future output of the system and compare them with the expected system output to obtain the cost function:
[0061] J=[F out (M,C state ,F control )-D out ] 2
[0062] Where J is the cost function, F out is the future output, M is the model, C state is the current state, F control is the future control variable, D out is the expected output.
[0063] The design and solution of MPC model predictive control usually adopts the quadratic programming method, as shown in the following formula:
[0064] minJ=Z T QZ+C T Z
[0065] The state equation of the discrete system is shown in the following formula: k+1 =AX k +BU k
[0066] Where B is the input matrix and A is the state transition matrix. The vector X defined k and U k As shown below:
[0067]
[0068] Where x(k+1|k) is the predicted system state at time k+1, where the prediction is made at time k, N is the maximum prediction space, i≤N. u(k+i|k) is the output of the predictive controller at time k+1, which is also predicted at time k, N-1 is the maximum prediction space, i≤N-1. The cost function is shown in the following formula:
[0069]
[0070] where x(k+i|k) T Qx(k+i|k) is the weighted sum of errors, u(k+i|k) T Ru(k+i|k) is the weighted sum of the inputs, x(k+N|k) TFx(k+N|k) is the terminal error (i.e., the error cost at the last step of the prediction time). Since x(k+N|k) is determined by u(k+N-1|k), u(k+N|k) is not needed, that is, there is no terminal input term. There are two variables in the expression of J (input u and state x). In order to eliminate the system state variable x and keep the control variable u:
[0071] x(k|k)=x k
[0072] x(k+1|k)=Ax(k|k)+Bu(k|k)=Ax k +Bu(k|k)
[0073] x(k+2|k)=Ax(k+1|k)+Bu(k+1|k)
[0074] =A 2 x k +ABu(k|k)+Bu(k+1|k)
[0075] …
[0076] x(k+N|k)=A N x k +A N-1 Bu(k|k)+…+Bu(k+N-1|k)
[0077] You can get:
[0078]
[0079] Define matrices M and C:
[0080]
[0081] X k =Mx k +CU k
[0082] The unfolded cost function is shown in the following formula:
[0083]
[0084] Defining the Matrix and As shown in the following formula:
[0085]
[0086] It turns out that:
[0087]
[0088] because and They are transposed and have the same value, that is:
[0089]
[0090] make:
[0091]
[0092] The final cost function obtained in the form of quadratic programming is shown in the following formula:
[0093]
[0094] The quadratic program can then be solved by adding system constraints.
[0095] Specifically, the path tracking problem solving process is to first establish the SSMR motion state equation as shown in the following formula:
[0096]
[0097] The state variable is z = [x, y, Ψ] T , z r =[x r ,y r ,Ψ r ] T is the reference trajectory, and the control variable is u=[v,w] T , the reference control variable is u r =[v r ,w r ] T .
[0098] The state space of SSMR is:
[0099]
[0100] At the target point z r The first-order Taylor expansion at gives:
[0101]
[0102] Depend on roll out:
[0103]
[0104] in:
[0105]
[0106] The kinematic model is discretized using the first-order difference method, and we can get:
[0107]
[0108] By changing the control variable u(k) to minimize
[0109]
[0110] in:
[0111]
[0112] The minimization objective function is shown in the following formula:
[0113]
[0114] where u t is the input vector, R is the input weight matrix, z t is the state vector, Q is the weight matrix of the state error, z r,t is the state vector of the reference trajectory, z r,T is the reference trajectory state vector at the final time step, Q f is the weight matrix of the state error at the final time step, z T is the trajectory state vector for the last time step.
[0115] The following constraints are met:
[0116] z0=z0
[0117] z t+1 =Az t +Bu t
[0118] u 0,t ≤speed max
[0119] u 0,t ≥speed min
[0120] u 1,t |≤w max
[0121] Where z0 is the initial state vector, matrices A and B are calculated according to the system discrete model, speed max and speed min are the upper and lower speed limits, w max is the upper limit of angular velocity.
[0122] The improved AST-MPC algorithm proposed in this application is to solve the potential problem that SSMR autonomous navigation has different requirements for sampling time in different situations such as straight lines, curves, obstacles, and external interference.
[0123] AST-MPC can dynamically adjust the sampling time according to the rate of change of the control input when the robot performs path tracking. For example, when the SSMR is traveling on a straight path, the curvature of the path and the input angular velocity change very little, and a longer sampling time can be used to reduce the computational burden. However, in paths with large changes in curvature, the sampling time has a great impact on the tracking performance because the input angular velocity also changes rapidly. Therefore, the AST-MPC algorithm improves the tracking performance of sections with large changes in control input in a targeted manner by gradually reducing the sampling time. This dynamic adjustment strategy enables the algorithm to adjust the sampling time more flexibly within the constraints and balance the relationship between computational efficiency and control performance.
[0124] Algorithm 1 is the pseudo code of AST-MPC.
[0125]
[0126]
[0127] The AST-MPC algorithm function is designed as follows:
[0128] The adaptive sampling time equation is shown in the following formula:
[0129] T(k+1)=T(k)+sign(ξ-C)·S
[0130] in:
[0131]
[0132] Specific experimental verification and comparative examples are given below.
[0133] Set α to 30, C to 0.6, S to 0.005, and consider energy consumption and safety issues. Set the robot's linear speed to 0.2 m / s. Figure 4 As shown, two curved paths are designed to simulate different path conditions in the robot navigation process, which are named Scenario 1 and Scenario 2. During the tracking process, the robot posture, tracking error, angular velocity, sampling time of each sampling point and other data under the AST-MPC algorithm of the present invention and the traditional MPC algorithm with fixed sampling time are derived, and the results after data processing are compared by data analysis software to evaluate the performance of the AST-MPC proposed in this application.
[0134] from Figure 5It can be seen that compared with the MPC algorithm with sampling time of 0.1s, 0.2s, and 0.25s, the tracking error of the AST-MPC algorithm proposed in this invention is significantly reduced when turning, and it is closer to the reference trajectory. Comparing the experimental results, the tracking performance of AST-MPC is better than that of MPC.
[0135] The adjustment of the sampling time in the AST-MPC algorithm is mainly determined by the rate of change of the angular velocity in the optimal control input calculated by MPC. This dynamic adjustment strategy ensures that the sampling time meets the different needs of the system dynamics and effectively responds to periods of drastic changes and weakening of the angular velocity. This adaptive sampling time makes the AST-MPC algorithm fully effective, enabling it to adjust its control strategy according to the specific dynamics of the system at different operating stages.
[0136] Furthermore, using Figure 6 The skid-steering mobile robot (SSMR) shown in FIG. is driven on a real path to conduct experiments to verify the effectiveness of the AST-MPC algorithm. The skid-steering mobile robot (SSMR) has four wheels, each driven by four independent motors, and the motion state of the robot is controlled by controlling the speed. It also has core components such as a depth camera, a laser radar, a touch screen, and an industrial personal computer. The parameters of the SSMR are shown in Table 1.
[0137] Table 1 Experimental platform parameters
[0138]
[0139] Further, the maximum error, minimum error and average error of different sampling times in scenario 1 are compared, and the results are shown in Table 2. By collecting data of different sampling times of the two algorithms and conducting comparative analysis, the tracking performance of the AST-MPC algorithm is demonstrated. Analysis of the data in Table 2 shows that compared with the MPC algorithm with a sampling time of 0.2s, the AST-MPC algorithm can reduce the maximum error by 37.2%, the minimum error by 33.9%, and the average error by 31.8%. For the MPC with a sampling time of 0.25s, the maximum error is reduced by 33.7%, the minimum error is reduced by 45.4%, and the average error is reduced by 34.4%. The analysis proves that AST-MPC has a significant improvement in tracking effect compared with MPC with different sampling times. Comparing AST-MPC with MPC with the same sampling time of 0.1s, the maximum error is almost the same. Although the minimum error of AST-MPC increases by 160.0%, the average error of AST, the error of MPC algorithm is reduced by 12.0%. The reason is that for MPC, the sampling time of 0.1s is already very small, and the tracking effect itself is good. It can be concluded that the tracking effect of the AST-MPC algorithm is not much different from that of the MPC with a sampling time of 0.1s. In summary, in addition to balancing the relationship between tracking effect and computing resources, AST-MPC can improve tracking performance.
[0140] Table 2 Tracking error for scenario 1
[0141] algorithm Maximum error Minimum error Average Error T=0.1MPC 0.02038 0.00030 0.01198 T=0.2MPC 0.03259 0.00118 0.01548 T=0.25MPC 0.03085 0.00143 0.01609 AST-MPC 0.02044 0.00078 0.01055
[0142] At the same time, the path tracking effects of the proposed MPC and AST-MPC algorithms under different sampling times in scenario 2 are compared. The sampling time setting and trajectory marking color are the same as those in scenario 1. In order to make an intuitive comparison, two segments with long sampling time and short sampling time are used to compare the tracking effects.
[0143] The results show that the proposed AST-MPC algorithm has significantly lower tracking error and is closer to the reference trajectory compared to the MPC algorithm with 0.2s and 0.25s sampling time. There is little difference in tracking performance compared to MPC and AST-MPC with a sampling time of 0.1s, but the overall performance of the AST-MPC algorithm is better. This observation emphasizes that the proposed algorithm improves the stability of control while maintaining tracking performance, resulting in better tracking results. See Table 3 for comprehensive details on the tracking error.
[0144] Table 3 Tracking error analysis table for scenario 2
[0145] algorithm Maximum error Minimum error Average Error T=0.1MPC 0.01915 0.00080 0.00748 T=0.2MPC 0.02640 0.00168 0.01531 T=0.25MPC 0.02970 0.00045 0.01848 AST-MPC 0.01230 0.00143 0.00663
[0146] Table 3 also compares the maximum error, minimum error, and average error. The experimental data analysis shows that compared with the MPC with a sampling time of 0.2s, the maximum error of the AST-MPC algorithm in this paper is significantly reduced by 53.4%, the minimum error is reduced by 14.9%, and the average error is reduced by 56.7%. Similarly, compared with the MPC with a sampling time of 0.25s, the maximum error of the AST-MPC algorithm is reduced by 58.5%, and although the minimum error increases by 217.8%, the average error is reduced by 64.1%. These findings confirm that the performance of AST-MPC is better than that of MPC with a longer sampling time, and the tracking effect is significantly enhanced. Compared with the MPC algorithm with a sampling time of 0.1s, the maximum error of AST-MPC is reduced by 35.8%, the minimum error is improved by 78.8%, but the average error is reduced by 11.3%. In summary, the proposed AST-MPC algorithm is an effective solution that effectively enhances the tracking performance while judiciously saving computing resources and achieving a good balance between tracking performance and control stability.
[0147] Furthermore, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of any of the aforementioned methods when executed by a processor.
[0148] Furthermore, the present invention provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of any of the aforementioned methods when executed by the processor.
[0149] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present invention.
[0150] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.
[0151] Although the present invention has been described in detail above with general description and specific embodiments, some modifications or improvements can be made on the basis of the present invention. The above description is only a preferred embodiment of the present invention and is not limited to the scope of the present invention. Other changes and modifications made by those skilled in the art without departing from the spirit and scope of protection of the present invention are still included in the scope of protection of the present invention.
Claims
1. A model predictive control method for a skid-steering mobile robot based on adaptive sampling time, characterized in that: The method comprises the following steps: Step S100: Initialize sampling time T; Step S200: By matching the current laser radar scan with the known map, the robot's position is calculated with the wheel speed meter, and the state variable is defined as P = (x, y, Ψ) T , control variable u=(v,ω), where the control variable is the velocity vector at the center of mass of the robot, v is the velocity, and ω is the angular velocity. The kinematic model of the wheeled robot is established, which is specifically expressed as follows: in is the velocity of the robot on the x-axis, is the component velocity of the robot on the y-axis, is the robot angular velocity; Step S300: discretize the robot kinematic model, set the predictive control model parameters including the prediction time domain, control time domain, initial sampling time, input weight matrix, state error weight matrix and constraint conditions, and construct a cost function including state error and control input; Step S400: Use the quadratic programming method to solve the predictive control model, minimize the cost function within the constraints to obtain the optimal control input sequence, and predict the state vector to be X(k+1)=A·X(k)+B·u(k), where X(k) is the state vector and u(k) is the control input vector. Update the state equation X(k)=X(k+1), where B is the input matrix and A is the state transfer matrix. Step S500: applying the first control input in the optimal control input sequence to the robot controller, storing the predicted state as the past state, and storing the control input as the past input, and using the calculated angular velocity change rate in the optimal control input sequence to calculate the sampling time T to be used when the vehicle moves next time; Step S600: Set T=T(k+1) to determine whether the calculated sampling time is within a preset threshold. If so, set it as a new sampling time. If it exceeds the maximum value of the threshold or is less than the minimum value of the threshold, set the sampling time to the maximum value or minimum value of the threshold accordingly. Step S700: Go to step S200 until the control ends.
2. The method according to claim 1, characterized in that The construction of the wheeled robot kinematic model in step S200 includes: The forward kinematics relationship of the wheeled robot is established, which is specifically expressed in the form of the following matrix product: Among them, V l and V r is the left and right wheel speed.
3. The method according to claim 1, characterized in that The acquisition process of the cost function in step S300 is: using the existing predictive control model, the current state variables of the system and the future control variables to predict the future output of the system, and comparing it with the expected system output to obtain the cost function: J=[F out (M,C state ,F control )-D out ] 2 Where J is the cost function, F out is the future output, M is the model, C state is the current state, F control is the future control variable, D out is the expected output.
4. The method according to claim 1, characterized in that The solution of the predictive control model in step S400 may further include: adding a speed upper limit and a lower limit, and an angular velocity upper limit as constraints for the quadratic programming solution.
5. The method according to claim 1, characterized in that Calculating the sampling time to be used for the next vehicle movement according to the angular velocity change rate in step S500 specifically includes: as the angular velocity change rate increases, the corresponding sampling time will be shorter, and as the angular velocity change rate decreases, the corresponding sampling time will be longer.
6. The method according to claim 5, characterized in that The specific formula for calculating the sampling time to be used when the vehicle moves next time is: T(k+1)=T(k)+sign(ξ-C)·S, where: α, S and C are constant parameters, which are set based on experience. α is the sensitivity of the rate of change of the angular velocity to the adjustment of the sampling time. S is the step size of the increase or decrease of the sampling time. Sign (ξ-C) is used to determine whether the sampling time should be increased or decreased in the next step. The sign function value 1, 0, -1 corresponds to an increase, no change and a decrease in the number of sampling times in the next step, respectively. is the rate of change of angular velocity in the control variable of the predictive control model.
7. The method according to claim 1, characterized in that The skid-steer mobile robot has four wheels, which are driven by four independent motors respectively, and the motion state of the robot is controlled by controlling the speed.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
