A path tracking control method and system based on feedforward model predictive control

Abstract

The application provides a path tracking control method and system based on feedforward model predictive control, relates to the motion control technical field of unmanned vehicles and mobile robots, and comprises the following steps: establishing a linear model predictive controller on an unmanned vehicle or a mobile robot; obtaining feedforward information of the unmanned vehicle or the mobile robot; constructing a feedforward model predictive controller based on the feedforward information; and combining the linear model predictive controller to perform path tracking control on the unmanned vehicle or the mobile robot. By introducing the feedforward information, the problem that the existing path tracking control method cannot effectively handle the influence of system constraints, cannot effectively utilize the front reference path information or has poor real-time performance can be solved, that is, the problem that the existing path tracking control method cannot effectively handle the influence of system constraints, cannot effectively utilize the front reference path information or has poor real-time performance can be solved.

Description

Technical Field

[0001] This invention relates to the field of motion control technology for unmanned vehicles and mobile robots, and in particular to a path tracking control method and system based on feedforward model predictive control. Background Technology

[0002] In existing path tracking control methods, those other than model predictive control struggle to effectively handle the effects of system constraints, making it difficult to achieve high-precision path tracking control under constraints such as front wheel steering angle and front wheel angular velocity. Among model predictive control methods, nonlinear model predictive control has the advantage of effectively utilizing forward reference path information and offers high accuracy, but its real-time performance is poor, making it difficult to deploy on hardware platforms with limited computing power. Linear model predictive control offers better real-time performance, but most research results employ control frameworks without a lead-in point, failing to achieve high-precision path tracking control under conditions where the reference path curvature amplitude is large and changes rapidly.

[0003] In summary, there is currently a lack of path tracking control methods that simultaneously possess the advantages of effectively handling system constraint effects, effectively utilizing forward reference path information, and having good real-time performance. Summary of the Invention

[0004] This invention provides a path tracking control method and system based on feedforward model predictive control, which solves the problem that existing path tracking control methods lack the three advantages of effectively handling system constraint effects, effectively utilizing forward reference path information, and having good real-time performance.

[0005] To achieve the aforementioned objectives, the present invention provides the following technical solution: a path tracking control method based on feedforward model predictive control, characterized in that the steps include:

[0006] S1. Establish a linear model predictive controller on unmanned vehicles or mobile robots;

[0007] S2. Obtain feedforward information from unmanned vehicles or mobile robots;

[0008] S3. Construct a feedforward model predictive controller based on feedforward information, and combine it with a linear model predictive controller to perform path tracking control for unmanned vehicles or mobile robots.

[0009] Preferably, in step S1, establishing a linear model predictive controller on the unmanned vehicle or mobile robot includes:

[0010] Obtain the kinematic or dynamic model of an unmanned vehicle or mobile robot;

[0011] Based on kinematic or dynamic models, obtain differential state prediction models of the controlled object;

[0012] A linear optimization function is preset, and then substituted into the differential state prediction model to obtain the standard quadratic form of linear control, thus establishing a linear model predictive controller.

[0013] Preferably, the linear control standard quadratic form includes:

[0014]

[0015] Where t is time t, and J is the objective function. H is the input increment matrix; H is the quadratic term matrix; G is the linear term matrix.

[0016] Preferably, in step S2, obtaining feedforward information from the autonomous vehicle or mobile robot includes:

[0017] Obtain reference path information for autonomous vehicles or mobile robots; the reference path information includes: curvature information of each reference path point of the autonomous vehicle or mobile robot;

[0018] The front wheel steering angle is calculated based on the reference path information.

[0019] Preferably, the calculation of the front wheel steering angle based on the reference path information includes:

[0020] Obtain curvature information;

[0021] Calculate the feedforward front wheel angle based on curvature information; set the reference path point where the curvature information is located as the aiming point; obtain the distance between each aiming point.

[0022] Preferably, in step S3, constructing a feedforward model predictive controller based on feedforward information includes:

[0023] Obtain feedforward information;

[0024] Based on feedforward information, a feedforward model predictive controller is constructed using a kinematic or dynamic model.

[0025] Preferably, in step S3, based on feedforward information, a feedforward model predictive controller is constructed using a kinematic model or a dynamic model, including:

[0026] Based on the curvature information at the preview point, the feedforward control input is represented by the feedforward front wheel angle;

[0027] A feedforward optimization objective function is preset, and the feedforward control input is substituted into the feedforward optimization objective function to obtain the standard quadratic form of the feedforward control, and a feedforward model predictive controller is constructed.

[0028] Preferably, the feedforward control standard quadratic form includes:

[0029]

[0030] Preferably, in step S3, the path tracking control of the unmanned vehicle or mobile robot is performed using a linear model predictive controller, including:

[0031] Obtain the standard quadratic form of linear control and the standard quadratic form of feedforward control;

[0032] As shown in the following formula (3), combine like terms of the linear control standard quadratic form and the feedforward control standard quadratic form:

[0033]

[0034] in:

[0035] By adding system constraints, we obtain the constrained optimization objective function as shown in the following formula (4);

[0036]

[0037] Solve the constrained optimization objective function; take the first value in the obtained control input sequence as the actual control input to complete path tracking control based on feedforward model predictive control.

[0038] A path tracking control system based on feedforward model predictive control, the system being used in the aforementioned path tracking control method based on feedforward model predictive control, the system comprising:

[0039] Linear control building blocks are used to build linear model predictive controllers on autonomous vehicles or mobile robots.

[0040] The information acquisition module acquires feedforward information from unmanned vehicles or mobile robots.

[0041] The path tracking control module is used to construct a feedforward model predictive controller based on the feedforward information, and combine it with the linear model predictive controller to perform path tracking control on the unmanned vehicle or mobile robot.

[0042] On the one hand, an electronic device is provided, the electronic device including a processor and a memory, the memory storing at least one instruction, the at least one instruction being loaded and executed by the processor to implement the above-described path tracking control method based on feedforward model predictive control.

[0043] On the one hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, the at least one instruction being loaded and executed by a processor to implement the above-described path tracking control method based on feedforward model predictive control.

[0044] The above technical solution has at least the following advantages compared with the existing technology:

[0045] The above-described solution, by introducing feedforward information, can solve the problem of the lack of a path tracking control method that simultaneously possesses three advantages: the ability to effectively handle the influence of system constraints, the ability to effectively utilize the preceding reference path information, and good real-time performance. In other words, it solves the problems of existing path tracking control methods that cannot effectively handle the influence of system constraints, cannot effectively utilize the preceding reference path information, or have poor real-time performance. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a schematic diagram of the path tracking control method based on feedforward model predictive control provided in an embodiment of the present invention;

[0048] Figure 2 This is a block diagram of a path tracking control system based on feedforward model predictive control provided in an embodiment of the present invention;

[0049] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0051] This invention addresses the problem that existing path tracking control methods lack the three advantages of effectively handling system constraint effects, effectively utilizing forward reference path information, and having good real-time performance. It provides a path tracking control method and system based on feedforward model predictive control.

[0052] like Figure 1 As shown, this embodiment of the invention provides a path tracking control method based on feedforward model predictive control, which can be implemented by an electronic device. Figure 1 The flowchart shown is for a path tracking control method based on feedforward model predictive control. The processing flow of this method may include the following steps:

[0053] S101. Establish a linear model predictive controller on unmanned vehicles or mobile robots;

[0054] In one feasible implementation, step S101, establishing a linear model predictive controller on the unmanned vehicle or mobile robot, includes:

[0055] Obtain the kinematic or dynamic model of an unmanned vehicle or mobile robot;

[0056] Based on kinematic or dynamic models, obtain differential state prediction models of the controlled object;

[0057] A linear optimization function is preset, and then substituted into the differential state prediction model to obtain the standard quadratic form of linear control, thus establishing a linear model predictive controller.

[0058] In one feasible implementation, the linear control standard quadratic form includes:

[0059]

[0060] Where t is time t, and J is the objective function. H is the input increment matrix; H is the quadratic term matrix; G is the linear term matrix.

[0061] In one feasible implementation, linear model predictive control can be based on the kinematic and dynamic models of autonomous vehicles and mobile robots. Taking an autonomous vehicle or mobile robot with Ackermann steering as an example:

[0062]

[0063] Where x and y are the horizontal and vertical coordinates of the rear axle center of the autonomous vehicle in the global coordinate system; θ is the heading angle of the autonomous vehicle in the global coordinate system; v is the longitudinal velocity; δ is the equivalent front wheel steering angle; and l is the wheelbase.

[0064] Generally, it can be abstracted as

[0065]

[0066] in:

[0067]

[0068] In one feasible implementation, the path is linearly expanded at the reference point closest to the center of the autonomous vehicle's rear axle:

[0069]

[0070] in:

[0071]

[0072] In this context, the variable with the subscript ref represents the information of the reference point, and the variable with the subscript 0 represents the current state information.

[0073] In each iteration cycle, the state variables basically conform to:

[0074]

[0075] Where t is time and T is the iteration period.

[0076] By combining equations (1-4) and (1-6), the prediction model for the differential state variables can be obtained.

[0077]

[0078] in:

[0079]

[0080] All differential state variables in the prediction time domain are:

[0081]

[0082] Where c is the number of control steps and p is the number of prediction steps.

[0083] Matrixing the differential state quantity prediction model:

[0084]

[0085] in:

[0086]

[0087] Then, the objective function can be designed as follows:

[0088]

[0089] Where Q is the weight matrix.

[0090] Substituting into the aforementioned differential state prediction model, and transforming formula (1-12) into a standard quadratic form:

[0091]

[0092] in:

[0093]

[0094] Linear model predictive control is equivalent to solving the following constrained standard quadratic form:

[0095]

[0096] Among them, the subscript li m It indicates the limit.

[0097] S102. Obtain feedforward information from unmanned vehicles or mobile robots;

[0098] In one feasible implementation, step S102, obtaining feedforward information from the unmanned vehicle or mobile robot, includes:

[0099] Obtain reference path information for autonomous vehicles or mobile robots; the reference path information includes: curvature information of each reference path point of the autonomous vehicle or mobile robot;

[0100] The front wheel steering angle is calculated based on the reference path information.

[0101] In one feasible implementation, the front wheel steering angle is calculated based on reference path information, including:

[0102] Obtain curvature information;

[0103] Calculate the feedforward front wheel angle based on curvature information; set the reference path point where the curvature information is located as the aiming point; obtain the distance between each aiming point.

[0104] S103. Construct a feedforward model predictive controller based on feedforward information, and combine it with a linear model predictive controller to perform path tracking control on unmanned vehicles or mobile robots.

[0105] In one feasible implementation, step S103, constructing a feedforward model predictive controller based on feedforward information, includes:

[0106] Obtain feedforward information;

[0107] Based on feedforward information, a feedforward model predictive controller is constructed using a kinematic or dynamic model.

[0108] In one feasible implementation, step S3 involves constructing a feedforward model predictive controller based on feedforward information using a kinematic or dynamic model, including:

[0109] Based on the curvature information at the preview point, the feedforward control input is represented by the feedforward front wheel angle;

[0110] A feedforward optimization objective function is preset, and the feedforward control input is substituted into the feedforward optimization objective function to obtain the standard quadratic form of the feedforward control, and a feedforward model predictive controller is constructed.

[0111] In one feasible implementation, considering that the curvature of some reference paths varies significantly, the feedforward front wheel steering angle can be calculated using curvature information in front of the autonomous vehicle. The reference path point containing the curvature information is the aiming point. The arc length between the aiming point and the reference point is the aiming distance, which satisfies the following condition:

[0112] d=kv (2-1)

[0113] Where d is the aiming distance and k is the adjustment coefficient.

[0114] The feedforward predictive control proposed in this patent can employ either a kinematic model or a dynamic model; the kinematic model will be used as an example here.

[0115] The desired yaw rate can be calculated from the velocity and curvature:

[0116]

[0117] Where κ is the curvature.

[0118] Substituting into the kinematic model, we can obtain the inverse kinematic model.

[0119] δ ref =arctan(κl) (2-3)

[0120] Since the curvature information at the aiming point is used, the feedforward control input, which is the feedforward front wheel steering angle, can be expressed as:

[0121] u pre =[δ pre (2-4)

[0122] In this context, the subscript 'pre' represents the aiming point parameter.

[0123] All feedforward control inputs can be expressed as:

[0124]

[0125] Matrix transformation yields:

[0126]

[0127] in:

[0128]

[0129] Since the goal of feedforward model predictive control is to make the control input close to the feedforward control input, the objective function can be designed as follows:

[0130]

[0131] in:

[0132]

[0133] R is the weight matrix.

[0134] In one feasible implementation, the feedforward control input obtained through iteration is substituted into the optimization objective function and transformed into a standard quadratic form:

[0135]

[0136] in:

[0137]

[0138] In one feasible implementation, step S103, combining a linear model predictive controller to perform path tracking control on the unmanned vehicle or mobile robot, includes:

[0139] Obtain the standard quadratic form of linear control and the standard quadratic form of feedforward control;

[0140] As shown in the following formula (3), combine like terms of the linear control standard quadratic form and the feedforward control standard quadratic form:

[0141]

[0142] in:

[0143] By adding system constraints, we obtain the constrained optimization objective function as shown in the following formula (4);

[0144]

[0145] Solve the constrained optimization objective function; take the first value in the obtained control input sequence as the actual control input to complete path tracking control based on feedforward model predictive control.

[0146] In one feasible implementation, the control method of this patent invention can be used for unmanned vehicles and mobile robots employing Ackerman steering and differential steering. Unmanned vehicles and mobile robots using the same steering type have the same kinematic model and the same dynamic model. When using different steering types, only the model inputs differ. The input for Ackerman steering is the equivalent front wheel steering angle, while the input for differential steering is the differential speed between the two wheels. There is a linear conversion relationship between the two. Furthermore, the model dimensions and outputs are the same, so the steering type does not affect the model processing. The inputs and outputs of the kinematic and dynamic models are also the same, and do not affect the model processing. In other words, whether it is an unmanned vehicle or a mobile robot, whether using a kinematic model or a dynamic model, the model can be abstracted into the form of formula (1-2), and the path tracking control method of this invention can be used.

[0147] In this embodiment of the invention, by introducing feedforward information, the problem of lacking a path tracking control method that simultaneously possesses the advantages of effectively handling system constraint influences, effectively utilizing forward reference path information, and having good real-time performance can be solved. That is, the problem of existing path tracking control methods being unable to effectively handle system constraint influences, unable to effectively utilize forward reference path information, or having poor real-time performance can be solved.

[0148] Figure 2 This is a schematic diagram of a path tracking control system based on feedforward model predictive control according to the present invention. The system 200 is used in the above-mentioned path tracking control method based on feedforward model predictive control, and the system 200 includes:

[0149] Linear control building block 210 is used to build linear model predictive controllers on unmanned vehicles or mobile robots;

[0150] Information acquisition module 220 is used to acquire feedforward information from unmanned vehicles or mobile robots;

[0151] The path tracking control module 230 is used to construct a feedforward model predictive controller based on the feedforward information, and combine the linear model predictive controller to perform path tracking control on the unmanned vehicle or mobile robot.

[0152] Preferably, the linear control building module 210 is used to obtain the kinematic or dynamic model of the unmanned vehicle or mobile robot;

[0153] Based on kinematic or dynamic models, obtain differential state prediction models of the controlled object;

[0154] A linear optimization function is preset, and then substituted into the differential state prediction model to obtain the standard quadratic form of linear control, thus establishing a linear model predictive controller.

[0155] Preferably, the linear control standard quadratic form includes:

[0156]

[0157] Where t is time t, and J is the objective function. H is the input increment matrix; H is the quadratic term matrix; G is the linear term matrix.

[0158] Preferably, the information acquisition module 220 is used to acquire reference path information of the unmanned vehicle or mobile robot; the reference path information includes: curvature information of each reference path point of the unmanned vehicle or mobile robot;

[0159] The front wheel steering angle is calculated based on the reference path information.

[0160] Preferably, the calculation of the front wheel steering angle based on the reference path information includes:

[0161] Obtain curvature information;

[0162] Calculate the feedforward front wheel angle based on curvature information; set the reference path point where the curvature information is located as the aiming point; obtain the distance between each aiming point.

[0163] Preferably, the path tracking control module 230 is used to acquire feedforward information;

[0164] Based on feedforward information, a feedforward model predictive controller is constructed using a kinematic or dynamic model.

[0165] Preferably, based on feedforward information, a feedforward model predictive controller is constructed using a kinematic or dynamic model, including:

[0166] Based on the curvature information at the preview point, the feedforward control input is represented by the feedforward front wheel angle;

[0167] A feedforward optimization objective function is preset, and the feedforward control input is substituted into the feedforward optimization objective function to obtain the standard quadratic form of the feedforward control, and a feedforward model predictive controller is constructed.

[0168] Preferably, the feedforward control standard quadratic form includes:

[0169]

[0170] Preferably, the path tracking control module 230 is used to acquire the linear control standard quadratic form and the feedforward control standard quadratic form;

[0171] As shown in the following formula (3), combine like terms of the linear control standard quadratic form and the feedforward control standard quadratic form:

[0172]

[0173] in:

[0174] By adding system constraints, we obtain the constrained optimization objective function as shown in the following formula (4);

[0175]

[0176] Solve the constrained optimization objective function; take the first value in the obtained control input sequence as the actual control input to complete path tracking control based on feedforward model predictive control.

[0177] In this embodiment of the invention, by introducing feedforward information, the problem of lacking a path tracking control method that simultaneously possesses the advantages of effectively handling system constraint influences, effectively utilizing forward reference path information, and having good real-time performance can be solved. That is, the problem of existing path tracking control methods being unable to effectively handle system constraint influences, unable to effectively utilize forward reference path information, or having poor real-time performance can be solved.

[0178] Figure 3 This is a schematic diagram of the structure of an electronic device 300 provided in an embodiment of the present invention. The electronic device 300 can vary considerably due to differences in configuration or performance. It may include one or more central processing units (CPUs) 301 and one or more memories 302. The memories 302 store at least one instruction, which is loaded and executed by the processors 301 to implement the steps of the path tracking control method based on feedforward model predictive control:

[0179] S1. Establish a linear model predictive controller on unmanned vehicles or mobile robots;

[0180] S2. Obtain feedforward information from unmanned vehicles or mobile robots;

[0181] S3. Construct a feedforward model predictive controller based on feedforward information, and combine it with a linear model predictive controller to perform path tracking control for unmanned vehicles or mobile robots.

[0182] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including instructions that can be executed by a processor in a terminal to perform the path tracking control method based on feedforward model predictive control described above. For example, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device, etc.

[0183] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

Claims

1. A path tracking control method based on feedforward model predictive control, characterized in that, The method steps include: S1. Establish a linear model predictive controller on unmanned vehicles or mobile robots; S2. Obtain feedforward information from unmanned vehicles or mobile robots; Obtain feedforward information from autonomous vehicles or mobile robots, including: Obtain reference path information for an autonomous vehicle or mobile robot; the reference path information includes: curvature information of each reference path point during the path tracking process of the autonomous vehicle or mobile robot; S3. Calculate the front wheel steering angle based on the reference path information; The calculation of the feedforward front wheel steering angle based on the reference path information includes: Set the aiming distance; take an aiming point on the reference path; obtain the curvature information of the aiming point; calculate the feedforward front wheel angle based on the curvature information; S4. Construct a feedforward model predictive controller based on the feedforward information, and combine it with the linear model predictive controller to perform path tracking control on the unmanned vehicle or mobile robot. S401. Construct a feedforward model prediction controller based on the feedforward information: Based on the feedforward information, a feedforward model predictive controller is constructed using a kinematic or dynamic model, including: By representing the feedforward control input through the front wheel steering angle, a feedforward control system is constructed. ; A feedforward optimization objective function is preset, and the feedforward control input is substituted into the feedforward optimization objective function to obtain the standard quadratic form of the feedforward control, and a feedforward model predictive controller is constructed. The standard quadratic form of feedforward control includes: （2） S402. Perform path tracking control on the unmanned vehicle or mobile robot using the linear model predictive controller: Obtain the standard quadratic form of linear control and the standard quadratic form of feedforward control; As shown in the following formula (3), combine like terms of the linear control standard quadratic form and the feedforward control standard quadratic form: （3） in: ; By adding system constraints, we obtain the constrained optimization objective function as shown in the following formula (4); (4) Solve the constrained optimization objective function; take the first value in the obtained control input sequence as the actual control input to complete the path tracking control based on feedforward model predictive control.

2. The method according to claim 1, characterized in that, In step S1, establishing a linear model predictive controller on the unmanned vehicle or mobile robot includes: Obtain the kinematic or dynamic model of an unmanned vehicle or mobile robot; Based on the kinematic or dynamic model, a differential state prediction model of the controlled object is obtained; A linear optimization function is preset, and the linear optimization function is substituted into the differential state prediction model to obtain a linear control standard quadratic form, thereby establishing a linear model predictive controller.

3. The method according to claim 2, characterized in that, The linear control standard quadratic form includes: （1） in, Let J be the time interval, and J be the objective function. H is the input increment matrix; H is the quadratic term matrix; G is the linear term matrix.

4. A path tracking control system based on feedforward model predictive control, characterized in that, The system is used in the path tracking control method based on feedforward model predictive control as described in any one of claims 1 to 3, the system comprising: Linear control building blocks are used to build linear model predictive controllers on autonomous vehicles or mobile robots. The information acquisition module acquires feedforward information from unmanned vehicles or mobile robots. The path tracking control module is used to construct a feedforward model predictive controller based on the feedforward information, and combine it with the linear model predictive controller to perform path tracking control on the unmanned vehicle or mobile robot.

Images