Integrated path planning and model predictive control method for multi-axle vehicle

Abstract

The application discloses a kind of multi-axle vehicle integration path planning and path tracking control method, belong to vehicle control engineering field;Specific process includes: step one gathers and collects vehicle information and environmental information, generates minimum turning radius path by solving constraint optimization problem;Step two is converted into the expected reference trajectory of vehicle by the path obtained through vehicle kinematics model;Step three is described as the optimal control problem of vehicle by the solution of control signal, and the expected torque sequence of each wheel is calculated by model predictive control;Step four is the first control amount in control sequence is input into vehicle, and the above process is repeated to carry out rolling horizon optimization.The application improves the adaptability of multi-axle vehicle to variable environmental conditions through rolling horizon optimization mechanism;At the same time, by designing model predictive control algorithm, the method for solving multivariable constraint optimization problem is used to calculate each wheel torque, which effectively improves the efficiency of torque distribution and ensures the path tracking effect.

Description

Technical Field

[0001] This invention belongs to the field of vehicle control engineering, specifically relating to an integrated multi-axle vehicle path planning and path tracking control method based on rolling time-domain optimization. Background Technology

[0002] Multi-axle distributed electric drive vehicles eliminate the mechanical transmission mechanisms found in traditional vehicles, offering high degrees of control freedom and fast motor response, resulting in higher power performance and better maneuverability. Simultaneously, with the large-scale application of autonomous driving technology, the unmanned operation of multi-axle special vehicles has become a development direction for military vehicles in the new era. Multi-axle vehicles can handle more complex working conditions and have significant application prospects in military transportation, special operations, and battlefield reconnaissance. However, harsh working conditions such as cliffs, precipices, earthen ridges, and shell craters necessitate higher demands on the path planning and dynamic control of multi-axle vehicles. The nonlinear dynamic characteristics and overdrive characteristics of multi-axle vehicles also make the design of corresponding path planning and control modules more complex.

[0003] Traditional vehicle lateral controllers use PID or LQR controllers designed based on bicycle models, relying on the continuous linearization of the planned path and often using linearized tire lateral force models to approximate vehicle dynamics. However, the complex terrain in multi-axle vehicle operating scenarios leads to stronger nonlinearities in the planned path, severely reducing control accuracy and potentially causing vehicle rollover in high-speed, small-turn-radius scenarios, seriously affecting vehicle safety. On the other hand, reasonable path planning is crucial for multi-axle vehicles operating in complex scenarios. Conventional autonomous vehicles mostly operate in urban or highway environments, where high-precision maps allow for accurate positioning, and path planning can be completed independently. In contrast, multi-axle vehicles typically operate on unstructured roads with varying objectives, frequently requiring fast and dynamic path planning. Traditional path planning methods cannot be directly applied to multi-axle vehicles. Therefore, in multi-axle vehicle applications, path planning and control must be considered as a whole. In summary, the new generation of special multi-axle distributed electric drive vehicles requires new path planning algorithms to cope with the changing and harsh operating environment, and new vehicle dynamics control methods to cope with the nonlinearity and dynamic constraints of the vehicle. The development of related algorithms has important practical significance for improving the lateral motion stability and driving safety of multi-axle vehicles.

[0004] Therefore, there is an urgent need to invent an integrated path planning and path tracking method for multi-axle vehicles to enhance their stability, effectively address challenges such as model nonlinearity and state constraints, and ultimately improve the driving safety of multi-axle vehicles. Summary of the Invention

[0005] In addressing the challenge that path planning and control algorithms in multi-axle intelligent vehicles are often independent and struggle to handle complex, dynamic environments, this invention proposes an integrated multi-axle vehicle path planning and tracking control method based on rolling time-domain optimization. First, the method collects vehicle and environmental information and generates a minimum turning radius path by solving a constrained optimization method. Then, the path obtained through the vehicle's kinematics model is transformed into the vehicle's desired reference trajectory. Next, the solution for the control signal is described as an optimal vehicle control problem, and the desired torque sequence for each wheel is calculated using model predictive control. Finally, the first control variable in the control sequence is input into the vehicle, and the above process is repeated for rolling time-domain optimization. By introducing an integrated approach into the planning and control of multi-axle vehicles, the adaptability of special vehicles to dynamic environments can be effectively improved, enhancing obstacle avoidance and driving smoothness. Introducing model predictive control into multi-axle vehicle control enhances the robustness of the vehicle control algorithm and effectively handles nonlinear dynamics and state constraints. The specific technical solution of this invention is described below.

[0006] See Figure 1 , Figure 7 As shown, the integrated multi-axle vehicle path planning and path tracking control method of the present invention includes the following steps:

[0007] Step 1: Generation of minimum curvature path based on rolling time-domain optimization;

[0008] Step 11: Collect surrounding environment information during vehicle movement using the SLAM algorithm and establish path constraints for the surrounding environment of multi-axle vehicles.

[0009] The surrounding environment path of a multi-axle vehicle can be measured by sensors (e.g., radar, cameras, lidar) and parameterized using two polygonal chains p1(x) and p2(x), where p1(x) represents the upper boundary of the path, p2(x) represents the lower boundary of the path, and x represents the position on the path. A schematic diagram of vehicle environmental perception and path planning is shown below. Figure 3 As shown.

[0010] In this invention, SLAM is an abbreviation for Simultaneous Localization and Mapping. Using sensors mounted on a multi-axle vehicle, a model of the vehicle's surrounding environment is built during the vehicle's movement, without prior information about the vehicle's motion environment, while simultaneously estimating the vehicle's motion.

[0011] Step 12: Parametricize the vehicle path using B-spline curves;

[0012] In this invention, the B-spline curve BS(x) for parameterized vehicle path planning is defined as:

[0013]

[0014]

[0015]

[0016] x represents the position on the path, which is also the parameter used in this invention for path planning using B-spline curves, and is also called the current position identifier.

[0017] n is the identifier of any control point. When n is 0 (n=0), it is the starting point of the path planning. For ease of explanation, n is also called the current control point. Control points before n are denoted as n-1 (i.e., the previous control point), and control points after n are denoted as n+1 (i.e., the next control point).

[0018] N is the total number of control points. The value of N is selected offline, and N+1≥k. k is the basis function S. n,k The function order of (x). The value of k is selected offline. For ease of explanation, k is also called the current function order. The function order before k is denoted as k-1 (i.e., the previous order), and the function order after k is denoted as k+1 (i.e., the next order).

[0019] Define the location node vector in path planning as T = (x0,...,x n-1 ,x n ,x n+1 ,...,x N+k x0 is the starting point of the path planning interval, x N+k The endpoint of the path planning interval is defined. The location node vectors are generated using a quasi-uniform method. n Let x be the position of control point n on the path planning. n-1 Let x be the position of control point n-1 on the path planning. n+1 Let x be the position of control point n+1 on the path planning. n+k-1 Let x be the position of control point n+k-1 on the path planning. n+k Let n+k be the position of control point n+k in the path planning.

[0020] α n Let n be the curve parameters for control point n.

[0021] S n,1 (x) is a first-order basis function of the control point n.

[0022] S n,k (x) is the k-th order basis function of the control point n.

[0023] S n,k-1 (x) is the k-1 basis function of the control point n.

[0024] S n+1,k-1 (x) is the k-1 basis function of the control point n+1.

[0025] Step 13: Design the path smoothness as the objective function, design an optimization problem with continuity constraints, and generate the path by solving the optimization problem;

[0026] In this invention, the designed path smoothness function F(x) is defined as:

[0027]

[0028] κ′(x) is the first derivative of κ(x).

[0029] S′(x) is the first derivative of S(x).

[0030] dx is the infinitesimal representation of the current position x.

[0031] The κ(x) mentioned above represents the curvature of the planned path, and its calculation method is as follows:

[0032]

[0033] S″(x) is the second derivative of S(x).

[0034] In this invention, the path continuity constraint is defined as follows:

[0035]

[0036] ε1 is the first adjustment parameter for path trajectory discontinuity.

[0037] ε2 is the second adjustment parameter for path trajectory discontinuity.

[0038] ε3 is the third adjustment parameter for path trajectory discontinuity.

[0039] S0(x0) is the path planned at the previous planning time step.

[0040] S(α 量 x) is the result of α 量 The required planned path after parameterization.

[0041] S′0(x0) is the first derivative of S0(x0).

[0042] S′(α 量 x0) is S(α) 量 The first derivative of x) via α 量 The required planned path after parameterization.

[0043] S″0(x0) is the second derivative of S0(x0).

[0044] S″(α 量 x0) is S(α) 量 The second derivative of x) via α 量 The required planned path after parameterization.

[0045] α 量 =[α1 ... α N [This represents the vector of decision variables for the curve parameters. α1 is the first decision parameter; α...] N for

[0046] The Nth decision parameter.

[0047] Furthermore, the optimization problem designed in this invention takes the form of:

[0048]

[0049] F(α 量 x) is the result of α 量 The parameterized path smoothness function.

[0050] The starting point x0 of the currently planned path is the second node in the path planned by the previous planned path S0(x).

[0051] Step 2: Modeling the path tracking error for multi-axle vehicles;

[0052] Step 21: Use a kinematic model to transform the generated path BS(x) into the desired path for a multi-axle vehicle, thus generating the desired path.

[0053] In this invention, the generated reference trajectory can be considered as being composed of the state vector x R =[x R y R γ R ] T Based on the sequence generated by the vehicle kinematics model, the vehicle kinematics model is as follows: Figure 4 As shown, the kinematic model is defined as follows:

[0054]

[0055] x R The desired X-axis position.

[0056] y R The desired Y-axis position.

[0057] γ R The desired vehicle yaw angle.

[0058] The desired X-axis velocity.

[0059] The desired Y-axis velocity.

[0060] Let yaw rate be the desired vehicle yaw rate.

[0061] u R The desired longitudinal speed.

[0062] v R The desired lateral speed.

[0063] r R Let be the desired yaw rate.

[0064] The conversion process is conducted using the following methods:

[0065]

[0066] ρ is the desired X-axis velocity selected offline.

[0067] t represents time.

[0068] S(x R () is based on the Y-axis path.

[0069] Where ρ > 0 represents the desired X-axis velocity selected offline. Based on the above formula, a corresponding reference trajectory can be generated.

[0070] Step 22: Based on the nonlinear dynamics model of a multi-axle vehicle, considering a 4-axle 8-wheel distributed model, taking into account the vehicle's lateral motion and yaw motion, the center of gravity sideslip angle, yaw angle, lateral tracking error, and yaw angle tracking error are selected as state variables, and the desired yaw moment is selected as the control variable to establish a path tracking error model.

[0071] See Figure 5 As shown, in the 4-axle 8-wheel distributed model, the wheels are labeled ij, where i represents the axle identifier and j represents the wheel identifier.

[0072] The first wheel on the left is marked as 11 (i.e., the left wheel on the first axle), the second wheel on the left is marked as 21 (i.e., the left wheel on the second axle), the third wheel on the left is marked as 31 (i.e., the left wheel on the third axle), and the fourth wheel on the left is marked as 41 (i.e., the left wheel on the fourth axle).

[0073] The first wheel on the right is marked as 12 (i.e., the right wheel on the first axle), the second wheel on the right is marked as 22 (i.e., the right wheel on the second axle), the third wheel on the right is marked as 32 (i.e., the right wheel on the third axle), and the fourth wheel on the right is marked as 42 (i.e., the right wheel on the fourth axle).

[0074] In this invention, the dynamic model of a multi-axle vehicle is as follows: Figure 5 As shown, its nonlinear dynamic model equations are established as follows:

[0075]

[0076]

[0077] Equation (10) is the vehicle's lateral motion equation, and equation (11) is the vehicle's yaw motion equation. Both equations consider small turning angle scenarios (δ1 and δ2 are relatively small, where δ1 is the front wheel axle angle and δ2 is the rear wheel axle angle).

[0078] β is the sideslip angle of the vehicle's center of gravity.

[0079] γ is the yaw rate of the vehicle.

[0080] v x This represents the longitudinal velocity.

[0081] v y This represents the lateral velocity.

[0082] l represents the wheel track.

[0083] I z Let be the moment of inertia about the z-axis of the center of mass.

[0084] L i Let be the distance from the i-th axis to the vehicle's center of mass.

[0085] m represents the vehicle's mass.

[0086] F x_ij This represents the longitudinal force of wheel ij.

[0087] F y_ij This represents the lateral force on wheel ij.

[0088] M z To add yaw moment.

[0089] The M mentioned z Defined as:

[0090]

[0091] F x_11 This is the longitudinal force of the first left wheel.

[0092] F x_12 This is the longitudinal force of the first right wheel.

[0093] F x_21 This is the longitudinal force of the second left wheel.

[0094] F x_22 This is the longitudinal force of the second right wheel.

[0095] F x_31 This is the longitudinal force of the third left wheel.

[0096] F x_32 This is the longitudinal force of the third right wheel.

[0097] F x_41 This is the longitudinal force of the fourth left wheel.

[0098] F x_42 This is the longitudinal force of the fourth right wheel.

[0099] Furthermore, define e y e represents the vehicle lateral tracking error. γ Let $\frac{ ...

[0100]

[0101]

[0102] This is the derivative of the vehicle lateral tracking error with respect to the vehicle's travel time.

[0103] This is the derivative of the vehicle sideslip angle error with respect to the vehicle's travel time.

[0104] v x The desired longitudinal speed.

[0105] β is the sideslip angle of the vehicle's center of gravity.

[0106] γ is the yaw rate of the vehicle.

[0107] κ(x) represents the curvature of the planned path.

[0108] Furthermore, a path tracking error system model is given, defining the error state variable as follows: The control variable u1 is the yaw moment M z The control variables u2 = [δ1 δ2] are the front wheel axle rotation angle and the front wheel axle rotation angle, respectively. The linear continuous time-invariant model is established as follows:

[0109]

[0110] A is the first matrix.

[0111] B is the second matrix.

[0112] E is the third matrix.

[0113] The matrices A, B, and E are defined as follows:

[0114]

[0115]

[0116]

[0117] K ij This represents the lateral stiffness of the tire, and its value can be approximated as a constant.

[0118] L i Let be the distance from the center of the i-th axis to the center of mass of the vehicle.

[0119] L1 is the distance from the center of the front wheel axle to the center of gravity of the vehicle.

[0120] L2 is the distance from the center of the rear wheel axle to the center of gravity of the vehicle.

[0121] K i1 This refers to the lateral stiffness of the front tire.

[0122] K i2 This refers to the lateral stiffness of the rear tire.

[0123] I z Let be the moment of inertia about the z-axis of the center of mass.

[0124] K 11 The lateral stiffness of tire 11.

[0125] K 12 The lateral stiffness of tire 12.

[0126] K 21 The lateral stiffness of tire 21.

[0127] K 22 This refers to the lateral stiffness of tire 22.

[0128] m represents the vehicle's mass.

[0129] v x The desired longitudinal speed.

[0130] Step 3: Optimization of expected yaw moment and multi-round moment allocation based on model predictive control;

[0131] Step 31: Design and optimize the objective function with the goal of minimizing the tracking error;

[0132] The design of the optimization problem requires the vehicle's current state information and the desired control objective for the path tracking error system. The overall control framework is as follows: Figure 2 As shown. Considering the vehicle's stability design requirements, the desired lateral motion should exhibit smooth steering characteristics, and the desired sideslip angle β is designed. ss =0,e y,d =0,e γ,d =0; Ignoring vertical load transfer, the desired yaw angle is Where L is the distance from the 1st axis to the 4th axis. The vector form of the expected value is defined as follows: Discretize the path tracking error system, design a sampling period of δt, and define the prediction step size as G. Then the prediction time is T = Nδt. Define x e (g|t) represents the predicted state of the discrete-time path tracking error system at time t after g steps of prediction. Then, the sequence X... e (t)=[x e (0|t) x e (1|t) ... x e [G|t)] represents all G-step states at time t, where x e (0|t)=x e (t) and x e (t) represents the sampled value at time t. Similarly, define the predictive control quantity u1(g|t) and the predictive control sequence U1(t) = [u1(0|t) u1(1|t) ... u1(G|t)]. The influence of vehicle steering is not considered during the prediction process, so let u2(g|t) = 0. The objective function of the optimization problem is defined as:

[0133]

[0134] x e,ss This represents the expected value of the vehicle trajectory.

[0135] e y,d This represents the expected value of the vehicle's lateral tracking error.

[0136] e γ,d This represents the expected value of the vehicle's sideslip angle error.

[0137] Q e It is the first weighted parameter matrix of 4×4 positive definiteness.

[0138] P e It is the second weighted parameter matrix, which is 4×4 positive definite.

[0139] R e These are weighted parameters.

[0140] In this invention, formula (19) is used to weight and adjust the predicted state, the terminal state and the control state respectively.

[0141] Step 32: Design the constraints in the optimization problem. Design the yaw moment distribution constraints based on the vertical load ratio of each axis. Design the drive wheel slip ratio constraints to prevent excessive slippage and instability of the drive wheels.

[0142] The constraint design includes two parts: state constraints and control constraints. The state constraints are composed of… Phase plane stability constraints, e y Lateral tracking error state constraints, e γ Yaw angle error state constraint; control constraint is M z Additional yaw moment constraint, ΔM z Yaw moment rate of change constraint. The specific form of the constraint is as follows:

[0143]

[0144] In the constraints, the subscript 'm' represents the minimum value of the corresponding variable, and the subscript 'M' represents the maximum value of the corresponding variable. The parameters are designed based on the actual physical structure parameters of the vehicle and the control requirements.

[0145] β is the sideslip angle of the vehicle's center of gravity.

[0146] Let be the time derivative of the vehicle's center of gravity sideslip angle.

[0147] B1 represents the boundary condition parameters under the stability domain, obtained through offline design based on vehicle information.

[0148] B2 represents the boundary condition parameters in the stability domain, obtained through offline design based on vehicle information.

[0149] e y,m This is for tracking the minimum dependent variable laterally.

[0150] e y,M To track the maximum dependent variable laterally.

[0151] e γ,m The minimum yaw angle is the dependent variable.

[0152] e γ,M The yaw angle is the maximum yaw rate.

[0153] M z,m The minimum strain of the additional yaw moment.

[0154] M z,M The strain is the maximum value of the additional yaw moment.

[0155] ΔM z,mThe minimum rate of change of yaw moment is the strain.

[0156] ΔM z,M The variable is the maximum rate of change of the yaw moment.

[0157] The final optimization problem is defined as:

[0158]

[0159] The decision variables for the optimization problem are U1(t)=[u1(0|t) u1(1|t) ... u1(G|t)], and the optimization result is defined as U1. * (t)=[u1 * (0|t) u1 * (1|t) ... u1 * [G|t)], where the superscript * represents the optimal value. M z * (t)=u1 * (0|t) represents the generated optimal expected yaw moment.

[0160] Step 33: Design the objective function and constraints for the torque distribution optimization problem to achieve torque distribution for each drive motor;

[0161] In this invention, the torque distribution method is designed based on an optimization algorithm, and its objective function is defined as:

[0162]

[0163] Where vector u torque =[T 11 T 12 T 21 T 22 T 31 T 32 T 41 T 42 ] T B represents the braking torque of each wheel ij, vector B torque Defined as r is the tire radius, W u W is the first weighted matrix. M M is the second weighting matrix. z * (t) represents the desired yaw moment, generated from the optimization problem in step 32.

[0164] The optimization problem takes the form of:

[0165]

[0166] Where F x,des This represents the expected value of the longitudinal force of the entire vehicle. U is the state transition matrix. torque,m U is the vector of minimum torque values ​​for each wheel. torque,M This is the vector of maximum torque values ​​for each wheel.

[0167] Step 4: Integrated multi-axle vehicle path planning and path tracking control;

[0168] In this invention, the path planning algorithm designed in step one is used to generate the desired trajectory for the optimization control algorithm designed in step three, and to generate control torques for each round. When the planned trajectory cannot meet the prediction time domain requirements of the control algorithm, path planning is performed again to generate a new trajectory for control.

[0169] Furthermore, the judgment relationship between the planned trajectory and the prediction time domain in the integrated path planning algorithm and path tracking control algorithm is as follows. Let D i The current environmental information is measured using sensors, and the step size between the first and second points of the planned path is M steps. The prediction step size in step three is N, requiring N < M. Let the number of iterations in the algorithm be k = 0, and the planned path use D... i The generated S i (x). Then when k≤MN, use S i Using (x) as the reference trajectory, the control algorithm in step three is used to generate the torque for each wheel, and the torque is applied to each tire. Each application increments the torque by 1, so k = k + 1. When k = M - N + 1, data D is re-acquired. i+1 Generate the desired path S using the newly collected data. i+1 (x); when k≤M, use S i (x) and S i+1 (x) Generate the corresponding trajectory, use the control algorithm in step three to generate the torque for each wheel, and apply the torque to each tire. Each time this is applied, k = k + 1. When k = M + 1, the current path point x... f Reset the value to x0, reset the iteration step number to k=0, and repeat the steps in S4 above until the target path endpoint is reached.

[0170] The advantages of the method of the present invention compared to the prior art are as follows:

[0171] (1) In view of the problem that the current path planning module and path tracking module are designed independently and it is difficult to handle complex and ever-changing scenarios, this invention proposes an integrated path planning and path tracking method, which improves the adaptability of multi-axle vehicles to changing environmental conditions through a rolling time-domain optimization mechanism.

[0172] (2) In view of the problems of nonlinear dynamic model of multi-axle vehicle, existence of state constraints and complex torque distribution, this invention designs a model predictive control algorithm and calculates the torque of each wheel by solving a multivariable constrained optimization problem, which effectively improves the efficiency of torque distribution and ensures the path tracking effect. Attached Figure Description

[0173] Figure 1 This is a schematic diagram of the integrated path planning and path tracking control method described in this invention.

[0174] Figure 2 This is a schematic diagram of the overall control block diagram of the multi-axle vehicle according to the present invention.

[0175] Figure 3 This is a schematic diagram of vehicle environment perception and path planning as described in this invention.

[0176] Figure 4 This is a schematic diagram of the kinematic model of the multi-axle vehicle described in this invention.

[0177] Figure 5 This is a schematic diagram of the multi-axle vehicle dynamics model described in this invention.

[0178] Figure 6 This is a schematic diagram of the multi-axle vehicle path tracking error model described in this invention.

[0179] Figure 7 A schematic diagram of the algorithm for the integrated path planning and path tracking control method described in this invention. Detailed Implementation

[0180] To enable those skilled in the art to better understand the present invention, the technical solution will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the 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.

[0181] See Figure 5 , Figure 6 As shown, in the 4-axis, 8-wheel distributed model, the first wheel on the left is labeled 11, the second wheel on the left is labeled 21, the third wheel on the left is labeled 31, and the fourth wheel on the left is labeled 41; the first wheel on the right is labeled 12, the second wheel on the right is labeled 22, the third wheel on the right is labeled 32, and the fourth wheel on the right is labeled 42.

[0182] This invention proposes an integrated path planning and path tracking control method for multi-axle vehicles based on rolling time-domain optimization. First, the method collects vehicle and environmental information and generates a minimum turning radius path by solving a constrained optimization problem. Then, the path obtained through the vehicle's kinematics model is transformed into the vehicle's desired reference trajectory. Next, the solution for the control signal is described as an optimal vehicle control problem, and the desired torque sequence of each wheel is calculated using model predictive control. Finally, the first control variable in the control sequence is input into the vehicle, and the above process is repeated for rolling time-domain optimization. By introducing the integrated approach into the planning and control of multi-axle vehicles, the adaptability of special vehicles to dynamic environments can be improved to a certain extent, enhancing obstacle avoidance and driving smoothness. By introducing model predictive control into multi-axle vehicle control, the robustness of the vehicle control algorithm can be enhanced, effectively handling nonlinear dynamics and state constraints. The overall process is as follows: Figure 1 , Figure 7 As shown, the overall control block diagram is as follows: Figure 2 As shown.

[0183] An integrated multi-axle vehicle path planning and path tracking control method includes the following steps:

[0184] Step 1: Generation of minimum curvature path based on rolling time-domain optimization;

[0185] Step 11: Collect information about the vehicle's surrounding environment using the SLAM algorithm and establish path constraints in the surrounding environment.

[0186] Step 12: Use B-spline curves to describe the path;

[0187] Step 13: Design the path smoothness as the objective function, design an optimization problem with continuity constraints, and generate the path by solving the optimization problem;

[0188] Step 2: Modeling the path tracking error for multi-axle vehicles;

[0189] Step 21: Use a kinematic model to transform the generated path into a desired path for a multi-axle vehicle, and generate the desired path;

[0190] Step 22: Based on the nonlinear dynamics model of a multi-axle vehicle, considering a 4-axle 8-wheel distributed model, taking into account the vehicle's lateral motion and yaw motion, the center of gravity sideslip angle, yaw angle, lateral tracking error, and yaw angle tracking error are selected as state variables, and the desired yaw moment is selected as the control variable to establish a path tracking error model.

[0191] Step 3: Optimization of expected yaw moment and multi-round moment allocation based on model predictive control;

[0192] Step 31: Design and optimize the objective function with the goal of minimizing the tracking error;

[0193] Step 32: Design the constraints in the optimization problem. Design the yaw moment distribution constraints based on the vertical load ratio of each axis. Design the drive wheel slip ratio constraints to prevent excessive slippage and instability of the drive wheels.

[0194] Step 33: Design the objective function and constraints for the torque distribution optimization problem to achieve torque distribution for each drive motor;

[0195] Step four: Use the path planning algorithm designed in step one to generate the desired trajectory for the optimized control algorithm designed in step three, and generate the control torque for each round; when the planned trajectory cannot meet the prediction time domain requirements of the control algorithm, re-plan the path and generate a new trajectory for control.

[0196] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for integrated path planning and path tracking control of multi-axle vehicles, characterized in that, include: Step 1: Collect vehicle and environmental information, and generate the minimum turning radius path by solving a constrained optimization problem; In the process of generating the minimum turning radius path based on rolling time-domain optimization, vehicle and environmental information are collected through the SLAM algorithm to establish vehicle-environment path constraints; then, B-spline curves are used to describe the minimum turning radius path; next, the smoothness of the minimum turning radius path is designed as the objective function, and the minimum turning radius path is generated by solving a constrained optimization problem. Step 2: The minimum turning radius path obtained through the multi-axle vehicle path tracking error model is transformed into the desired path for the multi-axle vehicle. The construction of a multi-axle vehicle path tracking error model includes: using the multi-axle vehicle path tracking error model to transform the generated minimum turning radius path into the desired path of the multi-axle vehicle, generating the desired path; then, based on the 4-axle 8-wheel distributed model, considering the vehicle's lateral motion and yaw motion, selecting the center of gravity sideslip angle, yaw angle, lateral tracking error, and yaw angle tracking error as state variables, and selecting the desired yaw moment as the control variable, to establish the multi-axle vehicle path tracking error model; Step 3: The desired path of the multi-axle vehicle is described as the optimal vehicle control problem. The desired torque sequence of each tire is calculated by optimizing the desired yaw moment of predictive control and the multi-wheel torque distribution model. The expected yaw moment optimization and multi-wheel torque distribution model based on predictive control includes: designing an optimization objective function with the goal of minimizing tracking error; then, designing yaw moment distribution constraints based on the vertical load ratio of each axle; designing drive wheel slip ratio constraints to prevent excessive slippage and instability of the drive wheels; and finally, obtaining the expected torque sequence of each tire. Step 4: Integrated path planning and path tracking control for multi-axle vehicles; The minimum turning radius path obtained in step one is used to generate the control torque of each tire; when the generated control torque of each tire cannot meet the rolling time domain optimization requirements, the multi-axle vehicle expected path tracking planning is re-performed to generate new control torque of each tire for control. The relationship between the control torque of each tire and the judgment of rolling time domain optimization: set up The current vehicle and environmental information is measured using sensors, where the step size between the first and second control points of the planned multi-axle vehicle's desired path is... Step 3 requires designing the total number of control points, i.e., the prediction step size. , ; Let the number of iterations be... The planned multi-axle vehicle desired path For use generate; when At that time, the planned multi-axle vehicle desired path As a reference trajectory, the torque of each tire is generated using the expected torque sequence of each tire obtained in step three, and the torque is applied to each tire, incrementing by 1 for each application. when At that time, data was re-collected. ,use Generate desired path for multi-axle vehicles ; when When using and Generate the desired path of the multi-axle vehicle at the next moment, use the desired torque sequence of each tire obtained in step 3 to generate the torque of each tire, and apply the torque to each tire, incrementing by 1 for each application; when At that time, the current path point Reset as the starting point of the path planning interval The number of iterations was reset to [number]. Repeat step four until you reach the end of the target path.

2. The integrated path planning and path tracking control method for multi-axle vehicles according to claim 1, characterized in that: B-spline curve in step one The definition of ,and and ; This represents the total number of control points. For any control point, the identifier number is used. Control points The curve parameters; Control points of basis functions of order; Control points First-order basis functions; The position on the path with the minimum turning radius; Control points Position on the path with the minimum turning radius; Control points Position on the path with the minimum turning radius; Control points of basis functions of order; Control points Position on the path with the minimum turning radius; Control points of basis functions of order; Control points Position on the path with the minimum turning radius.

3. The integrated path planning and path tracking control method for multi-axle vehicles according to claim 2, characterized in that: Minimum turning radius path smoothness function in step one Defined as The aforementioned The curvature of the path representing the minimum turning radius is ; The endpoint of the path with the minimum turning radius; The starting point of the path with the minimum turning radius; for The first derivative; for The first derivative; for The second derivative of .

4. The integrated path planning and path tracking control method for multi-axle vehicles according to claim 3, characterized in that: The constrained optimization problem solved in step one is as follows: ; The first adjustment parameter for the path discontinuity at the minimum turning radius; The second adjustment parameter for the discontinuous path of the minimum turning radius; The third adjustment parameter for the discontinuous path at the minimum turning radius; The path with the minimum turning radius at the previous moment; For the The parameterized path with the minimum turning radius; To The first derivative; for The first derivative The parameterized path with the minimum turning radius; To The second derivative; for The second derivative of The parameterized path with the minimum turning radius; The vector of decision variables for curve parameters; This is the first decision parameter; For the first One decision parameter; For the The parameterized minimum turning radius path smoothness function; The upper boundary of the path with the minimum turning radius; This represents the lower boundary of the path with the minimum turning radius.

Images