A novel integrated redundancy control method for path following and yaw stability chassis

Abstract

The application discloses a novel path tracking and yaw stability chassis integrated redundancy control method, relates to the path tracking control algorithm technical field of unmanned racing cars, and aims at solving the problems that the existing MPC algorithm of the nonlinear dynamics model consumes large calculation resources and is difficult to meet the real-time requirement of the algorithm. The method comprises the following steps: 1, a path tracking control algorithm is predicted by using an upper model; a prediction model of the model prediction algorithm is a 5-degree-of-freedom particle nonlinear vehicle dynamics model; 2, a bottom chassis optimization control strategy is designed; 2-1, in the first layer controller, the longitudinal tire force of each wheel and the front and rear wheel steering angle are optimized and distributed; 2-2, in the second layer controller, the ideal longitudinal tire force of each wheel is distributed to the specific independent motor torque or the brake torque of the hydraulic brake through a distribution strategy based on a simple distribution rule.

Description

Technical Field

[0001] This invention relates to the field of path tracking control algorithm technology for unmanned racing cars, specifically a novel path tracking and yaw stability chassis integrated redundancy control method. Background Technology

[0002] In the current field of vehicle control, chassis integrated optimization control allocation strategies have seen rapid development in passenger vehicle stability and safety control. Especially for distributed drive independent steering electric vehicles, the distributed independent drive motors and independent steering actuators provide additional control redundancy. Therefore, by systematically designing chassis optimization allocation strategies, the steering angle, drive, and braking torque of the chassis can be optimized, improving vehicle stability control. Currently, chassis integrated control is mostly used in passenger vehicle design and development, lacking development and design specifically for racing cars. Racing cars aim for maximum speed on the track; therefore, under high-speed extreme cornering conditions, chassis optimization allocation control strategies can maximize the friction limit of each tire, unleash the tire's potential, generate ideal yaw moment, and significantly improve the steering performance and stability of racing cars during extreme cornering.

[0003] On the other hand, many universities and enterprises have developed path tracking controllers for autonomous vehicles. Currently, the most advanced control strategy employs model predictive control (MMC) algorithms for steering control. MMC algorithms can predict the vehicle's kinematics and dynamics in advance, improving vehicle control compared to traditional feedback control. However, most current autonomous vehicle path tracking controller designs use independent modules. The vehicle's steering angle, optimized by the path tracking controller, along with throttle and brake pedal opening signals from the drive-by-wire and braking systems, are transmitted to the vehicle chassis. The chassis then controls the vehicle through corresponding steering controllers, drive motor controllers, and brakes. There is a lack of a comprehensive, integrated control strategy that combines upper-level path tracking control with chassis-integrated optimization control. Especially for autonomous electric racing cars with distributed drive motors, four-wheel independent steering, and independent brakes, a comprehensive control allocation strategy is needed to optimize the allocation of independent steering angles, drive motors, and braking torque, ensuring path tracking accuracy and vehicle dynamic stability at maximum speeds.

[0004] At present, MPC algorithm is generally used to solve path tracking problem, and MPC algorithm needs accurate nonlinear complex dynamic model to maximize the advantage of MPC algorithm, especially in high-speed limit race road conditions, the nonlinear of vehicle dynamics and tire model should be considered. However, the MPC algorithm based on complex nonlinear dynamic model greatly consumes computing resources, and it is difficult to meet the real-time requirements of the algorithm. Therefore, it is particularly important to propose a method which can maximize the potential of MPC algorithm to improve path tracking performance and ensure the real-time effectiveness of the algorithm, so it does not meet the existing demand, and a new type of path tracking and yaw stability chassis integrated redundancy control method is proposed. SUMMARY

[0005] The purpose of the present application is to provide a new type of path tracking and yaw stability chassis integrated redundancy control method to solve the problem of large consumption of computing resources of MPC algorithm based on nonlinear dynamic model and difficulty in meeting the real-time requirements of the algorithm in the background technology.

[0006] To achieve the above purpose, the present application provides the following technical scheme: a new type of path tracking and yaw stability chassis integrated redundancy control method, comprising the following steps: 1, using an upper model to predict a path tracking control algorithm, the prediction model of the model prediction algorithm is a 5-degree-of-freedom particle nonlinear vehicle dynamics model, which includes vehicle longitudinal velocity, lateral velocity, yaw angular velocity, path tracking yaw angle deviation and lateral displacement deviation;

[0007] Step 2, the bottom layer chassis optimization control strategy design adopts two-layer hierarchical design;

[0008] Step 2-1, in the first layer controller, based on the ideal total longitudinal tire force, total lateral tire force and ideal yaw moment optimized by the upper path tracking controller, the longitudinal tire force of each wheel and the front and rear wheel steering angle are optimized and distributed;

[0009] Step 2-2, in the second layer controller, the ideal longitudinal tire force of each wheel is distributed to the specific independent motor torque or hydraulic brake brake torque through the distribution strategy based on the simple distribution rule.

[0010] Preferably, the discretized prediction model in step 1 is as follows:

[0011]

[0012]

[0013]

[0014] F xd,k+1 =F xd,k +Ts F xd

[0015] F yd,k+1 = F yd,k + T s F yd

[0016] M zd,k+1 = M zd,k + T s M zd

[0017] ψ e,k+1 = ψ e,k + T s (r k - p 1,k v x,k )

[0018] Y e,k+1 = Y e,k + T s (v x,k sin ψ e,k + v y,k cos ψ e,k )

[0019] where v x，k , v y,k , r k denote the longitudinal velocity, lateral velocity and yaw rate value at discrete time k, T s denotes the discrete time step, F xd,k , F yd,k , M zd,k are the ideal total longitudinal tire force, total lateral tire force and yaw moment at k, m denotes the vehicle mass, I z denotes the vehicle moment of inertia about the vertical axis, ψ e,k is the yaw angle error at k, Y e，k is the lateral offset error at k while tracking, P 1,k denotes the path curvature at k, in order to eliminate the chattering of the optimization output, unlike the total lateral tire force, total longitudinal tire force and yaw moment used as the optimization output in the traditional model predictive method, the first derivative of the total lateral force, total longitudinal force and yaw moment is used as the optimization output;

[0020] The optimization objective function of the model predictive algorithm includes minimizing the velocity tracking error, minimizing the path tracking error and penalizing the optimization control output, which can be specifically represented by the following formula:

[0021]

[0022] where v xd，ku k denotes the ideal speed of the vehicle at time k, u xd,k yd,k zd,k k denotes the control output of the vehicle, Q v ψ Y are the optimization objective weight coefficients, corresponding to minimizing the speed tracking error, minimizing the yaw angle tracking error and minimizing the lateral displacement tracking error, respectively, Q u correspond to the weight coefficients for penalizing the optimization output control, Q s1 and Q s2 are the weight coefficients for soft constraints;

[0023] The constraint conditions of the model predictive algorithm should first consider the friction limit of the vehicle tire, i.e. the total lateral tire force and the total longitudinal tire force of the optimization output should satisfy the requirement of the friction ellipse, and the yaw moment of the optimization output should also satisfy the upper and lower limits, which can be specifically expressed by the following formula:

[0024]

[0025]

[0026]

[0027] wherein μ denotes the tire road friction coefficient, g is the gravity acceleration, M zmax is the upper limit of the yaw moment, M zmin is the lower limit of the yaw moment, s1 and s2 are soft constraints in the constraint conditions, and the soft constraints are set here to prevent the optimization process from exceeding the constraint boundary and thus obtaining invalid solutions, and the relaxation and tightening of the soft constraints can be achieved by adjusting the weight coefficients Q s1 and Q s2 in the optimization objective function.

[0028] Preferably, the first layer optimization allocation strategy in step 2-1, the optimization objective function can be expressed by the following formula:

[0029]

[0030] wherein u T =[u1 u2 u3 u4 u5 u6] denotes the left front wheel longitudinal tire force, the right front wheel longitudinal tire force, the left rear wheel longitudinal tire force, the right rear wheel longitudinal tire force, the front wheel steering angle and the rear wheel steering angle 6 optimization control output values;

[0031] is a constant matrix,

[0032] ​​​​​ The matrix contains the ideal total tire force F of the upper path tracking controller xd , F yd and the ideal yaw moment M zd , Q1 = diag [q 11 ; q 12 ; q 13 ] q11 is the weight coefficient of the ideal total longitudinal tire force, q12 is the weight coefficient of the ideal total lateral tire force, and q13 is the weight coefficient of the ideal yaw moment, Q2 = diag [q 21 ; q 22 ; q 23 ; q 24 ; q 25 ; q 26 ] corresponds to the penalty function coefficient of the left front wheel longitudinal tire force, the right front wheel longitudinal tire force, the left rear wheel longitudinal tire force, the right rear wheel longitudinal tire force, the front wheel steering angle, and the rear wheel steering angle, respectively;

[0033] The optimization objective function also needs to meet the following friction limit boundary conditions:

[0034]

[0035]

[0036]

[0037]

[0038] Wherein F zrl , F zrr are the vertical loads of each wheel, a fld , a frd , a rld , a rrd are the side slip angles of each wheel, in addition, each wheel steering angle should also meet the steering angle physical limit constraint, written as a quadratic programming form based on quadratic constraints, which can improve the operation efficiency of the optimization algorithm.

[0039] Preferably, the second layer control distribution strategy in step 2-2 specifically distributes each wheel ideal longitudinal tire force obtained by the first layer optimization to the torque of each wheel independent drive motor and brake system, in order to optimize energy loss, when specifically distributing the longitudinal tire force of each wheel, the independent drive motor is preferred to be used to generate brake torque by reversing, and when the motor reverse torque is insufficient, the brake torque generated by the hydraulic independent brake is used for compensation.

[0040] Compared with the prior art, the beneficial effects of the present application are:

[0041] 1. The path tracking and chassis integrated comprehensive control strategy of the application comprises an upper model prediction path tracking controller and a lower chassis optimization control distribution strategy, the upper path tracking controller tracks the ideal path on the race track and the ideal maximum vehicle speed, and sends the total tire force and yaw moment required by the upper controller to the lower chassis controller, and the lower chassis optimization controller optimizes the steering angle of the independent steering motor, the torque of the independent drive motor and the independent brake torque based on the ideal force and moment required by the upper layer, the two-layer structure can simplify the prediction model in the upper model prediction algorithm to a particle model, and the lower chassis control distribution strategy considers the complex vehicle dynamics model and tire model, thereby improving the running efficiency of the integrated algorithm.

[0042] 2. The two-stage path tracking controller designed by the research scheme can make full use of the additional yaw moment generated by the distributed drive independent motor and the front and rear wheel active steering to improve the path tracking performance and yaw stability of the vehicle through real-time simulation verification, and the racing car can approach the friction limit of the tire when turning at a small radius, achieving a high speed while ensuring the accuracy of path tracking. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 The total block diagram of the controller of the application is designed;

[0044] Figure 2 The second layer control distribution strategy logic block diagram in the lower chassis control strategy of the application is designed;

[0045] Figure 3 The simulation effect verification diagram of the two-stage path tracking controller of the application is designed;

[0046] Figure 4 The robustness analysis diagram of the control effect of the two-stage path tracking controller of the application is designed;

[0047] Figure 5 The real-time analysis diagram of the control algorithm of the application is designed. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all.

[0049] Please refer to Figures 1-5In one embodiment, the application provides a novel integrated redundancy control method for path tracking and yaw stability chassis, comprising the following steps: 1. using an upper model to predict a path tracking control algorithm, the prediction model of the model prediction algorithm is a 5-degree-of-freedom particle nonlinear vehicle dynamics model, including vehicle longitudinal velocity, lateral velocity, yaw angular velocity, path tracking yaw angle deviation and lateral displacement deviation; 2. designing a bottom chassis optimization control strategy, adopting a two-layer hierarchical design; 2-1. in the first layer controller, based on the ideal total longitudinal tire force, total lateral tire force and ideal yaw moment obtained by the upper path tracking controller optimization, the longitudinal tire force of each wheel and the front and rear wheel steering angle are optimized and distributed; 2-2. in the second layer controller, the ideal longitudinal tire force of each wheel is distributed to the specific independent motor torque or hydraulic brake brake torque through a distribution strategy based on a simple distribution rule.

[0050] Further, the discretized prediction model in step 1 is as follows:

[0051]

[0052]

[0053]

[0054] F xd,k+1 =F xd,k +T s F xd

[0055] F yd,k+1 =F yd,k +T s F yd

[0056] M zd,k+1 =M zd,k +T s M zd

[0057] ψ e,k+1 =ψ e,k +T s (r k -p 1,k v x,k )

[0058] Y e,k+1 =Y e,k +T s (v x,k sinψ e,k +v y,k cosψ e,k )

[0059] wherein vx，k , v y,k , r k denote the longitudinal velocity, lateral velocity and yaw rate value at discrete time k, T s denotes the discrete time step, F xd,k , F yd,k , M zd,k are the ideal total longitudinal tire force, total lateral tire force and yaw moment at k, m denotes the vehicle mass, I z denotes the vehicle moment of inertia about the vertical axis, ψ e,k is the yaw angle error at k, Y e，k is the lateral offset error at k, P 1,k denotes the path curvature at k, in order to eliminate the chattering of the optimization output, unlike the total lateral tire force, total longitudinal tire force and yaw moment used as the optimization output in the traditional model predictive method, the first derivative of the total lateral force, total longitudinal force and yaw moment is used as the optimization output;

[0060] The optimization objective function of the model predictive algorithm includes minimizing the velocity tracking error, minimizing the path tracking error and penalizing the optimization control output, which can be specifically represented by the following formula:

[0061]

[0062] where v xd， k denotes the ideal velocity of the vehicle at k, u k = [F xd,k ; F yd,k ; M zd,k ] k denotes the control output of the vehicle, Q v , Q ψ , Q Y are the optimization objective weight coefficients, corresponding to minimizing the velocity tracking error, minimizing the yaw angle tracking error and minimizing the lateral offset tracking error, respectively, Q u corresponds to the weight coefficient of penalizing the optimization output control, Q s1 and Q s2 are the weight coefficients of soft constraints;

[0063] The constraint condition of the model predictive algorithm should first consider the friction limit of the vehicle tire, that is, the total lateral tire force and the total longitudinal tire force of the optimization output should meet the requirement of the friction ellipse, and the yaw moment of the optimization output should also meet the upper and lower limits, which can be specifically represented by the following formula:

[0064]

[0065]

[0066]

[0067] where μ denotes the tire-road friction coefficient, g is the gravity acceleration, M zmax is the upper limit of the yaw moment, M zmin is the lower limit of the yaw moment, s1 and s2 are soft constraints in the constraint conditions, which are set to prevent the optimization process from exceeding the constraint boundary and obtaining invalid solutions, and the relaxation and tightening of the soft constraints can be achieved by adjusting the weight coefficients Q s1 and Q s2 .

[0068] Further, the optimization distribution strategy in step 2-1 can be represented by the following optimization objective function:

[0069]

[0070] where u T =[u1 u2 u3 u4 u5 u6] represents the six optimization control output values of the front left wheel longitudinal tire force, the front right wheel longitudinal tire force, the rear left wheel longitudinal tire force, the rear right wheel longitudinal tire force, the front wheel steering angle, and the rear wheel steering angle;

[0071] is a constant matrix,

[0072] The matrix contains the ideal total tire force F xd , the ideal lateral tire force F yd , and the ideal yaw moment M zd , Q1 = diag[q 11 ; q 12 ; q 13 ] is the weight coefficient of the ideal total longitudinal tire force, the weight coefficient of the ideal total lateral tire force, and the weight coefficient of the ideal yaw moment, Q2 = diag[q 21 ; q 22 ; q 23 ; q 24 ; q 25 ; q 26 ] respectively correspond to the penalty function coefficient of the front left wheel longitudinal tire force, the front right wheel longitudinal tire force, the rear left wheel longitudinal tire force, the rear right wheel longitudinal tire force, the front wheel steering angle, and the rear wheel steering angle;

[0073] The optimization objective function also needs to satisfy the following friction limit boundary conditions:

[0074]

[0075]

[0076]

[0077]

[0078] wherein F zrl ,F zrr are the vertical load of each wheel, a fld ,a frd ,a rld ,a rrd are the side slip angle of each wheel, in addition, each wheel steering angle should also satisfy the steering angle physical limit constraint, written as a quadratic programming form based on the quadratic constraint, which can improve the operation efficiency of the optimization algorithm.

[0079] Further, the second layer control distribution strategy in step 2-2 specifically distributes each wheel ideal longitudinal tire force obtained by the first layer optimization to the torque of each wheel independent drive motor and brake system, in order to optimize energy loss, when distributing the longitudinal tire force of each wheel, the use of independent drive motor reverse rotation to generate brake torque is given priority, and when the motor reverse torque is insufficient, the brake torque generated by the hydraulic independent brake is used to compensate, in Figure 2 j u j,min represents the maximum negative torque of the motor; R ω represents the tire radius; T mj ,T bj represent the allocated motor output torque and hydraulic brake torque, respectively.

[0080] Application

[0081] The following simulation results are based on high-fidelity IPG Carmker simulation software, simulating a section of the Silverstone race track in the UK. Figure 3 ​(left) shows the ideal track of the path following. The chassis structure of the autonomous racing car in simulation contains front-wheel active steering and rear-wheel active steering. A driving motor is installed in the front axle and the driving torque of the motor is equally distributed to the left and right wheels through a differential. Independent driving motors are installed in the rear wheels, and hydraulic brakes are installed in the front axle and the rear axle, respectively. The simulation results compare the control effects of four different path following controls: 'MPC+CA' refers to the designed two-stage path following controller; 'MPC+CAwithoutTV' refers to the designed two-stage path following controller in which the two rear wheels produce the same driving or braking torque and cannot produce additional yaw moment; 'MPC+CAwithoutRS' refers to the designed two-stage path following controller in which the rear-wheel active steering function is removed; and 'FB+CA' refers to the traditional error feedback path following control algorithm used in the upper path following control instead of the designed upper model predictive path following algorithm. Figure 3 (middle) shows that the path following error of the designed 'MPC+CA' path following control strategy is significantly smaller than that of the traditional 'FB+CA'. At the same time, the path following performance of 'MPC+CA' is better than that of 'MPC+CAwithoutTV' and 'MPC+CAwithoutRS', fully embodying the improvement of the rear-wheel independent driving motor and the rear-wheel active steering on the path following control. Similarly, Figure 3 (right) also shows that the designed 'MPC+CA' can achieve better ideal speed tracking effect than other control strategies, and is closest to the ideal speed limit.

[0082] Figure 4 Robustness analysis of the designed controller is shown. First, the inaccurate estimation of the tire road friction coefficient is considered (MPC+CAmiu=0.7), which assumes that the friction coefficient is incorrectly estimated as 0.7 (the true value is 1). Second, the random white noise of acceleration and yaw rate is considered (MPC+CA+noise), Figure 4 The simulation results of the path following error and the speed tracking error in the middle show the good robustness of the designed controller.

[0083] Real-time analysis of the path following control algorithm is very important. Only by verifying the real-time of the algorithm can the real-time online optimization of the control algorithm in the real car be ensured. Figure 5 The optimal solution solving time of the optimization control algorithm at each discrete time step is shown. In the simulation, the discrete time step is assumed to be 50 ms, and the maximum value of the optimal solution solving time is not more than 30 ms, so the real-time of the algorithm can be effectively verified.

[0084] It will be apparent to those skilled in the art that the application is not limited to the details of the above-exemplified embodiments and that the present application can be implemented in other particular forms without departing from the spirit or essential characteristics of the present application. The embodiments should therefore be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the above description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. No reference signs in the claims should be considered as limiting the scope of the claims with respect to the figures of the patent document.

Claims

1. A novel path tracking and roll stability chassis integrated redundancy control method, wherein, Comprise: Step 1, using upper model prediction path tracking control algorithm, the prediction model of model prediction algorithm is a 5-DOF particle nonlinear vehicle dynamics model, including vehicle longitudinal velocity, lateral velocity, yaw rate, path tracking yaw angle deviation and lateral displacement deviation; Step 2, the bottom chassis optimization control strategy design, using two-layer hierarchical design; Step 2-1, in the first layer controller, based on the ideal total longitudinal tire force, total lateral tire force and ideal yaw moment obtained by the upper path tracking controller optimization, the longitudinal tire force of each wheel and the front and rear wheel steering angle are optimized and distributed; Step 2-2, in the second layer controller, the ideal longitudinal tire force of each wheel is distributed to the specific independent motor torque or hydraulic brake torque through the distribution strategy based on the simple distribution rule; The discretized prediction model in step 1 is as follows: wherein , , respectively represent longitudinal velocity, lateral velocity and yaw rate value at discrete time k, denotes discrete time step, , is ideal total longitudinal tire force, total lateral tire force and yaw moment at k, m represents vehicle mass, denotes vehicle moment of inertia along vertical direction, is yaw angle error at k, is lateral offset error at k when tracking, denotes path curvature at k, in order to eliminate the chattering of optimization output, unlike the total lateral tire force, total longitudinal tire force and yaw moment used as optimization output in the traditional model prediction method, the first derivative of total lateral force, total longitudinal force and yaw moment is used as optimization output. The optimization objective function of model prediction algorithm includes minimizing the speed tracking error, minimizing the path tracking error and punishing the optimization control output, which can be represented by the following formula: wherein denotes the ideal speed of the vehicle at time k, denotes the control output of the vehicle, are optimization target weight coefficients, corresponding to minimizing the speed tracking error, minimizing the yaw angle tracking error and minimizing the lateral offset tracking error, respectively, correspond to weight coefficients penalizing the optimization output control quantities, and is a weight coefficient of the soft constraint; The constraint condition of model prediction algorithm should first consider the friction limit of vehicle tire, that is, the total lateral tire force and total longitudinal tire force of optimization output should meet the requirement of friction ellipse, and the yaw moment of optimization output should also meet the upper and lower limits, which can be represented by the following formula: where µ represents the tire road friction coefficient, g is the gravity acceleration, is the upper limit of the yaw moment, is the lower limit of the yaw moment, s1 and s2 are soft constraints in the constraint conditions, the soft constraints are set here to prevent the optimization process from exceeding the constraint boundary to obtain invalid solutions, and the relaxation and tightening of the soft constraints can be adjusted by adjusting the weight coefficient in the optimization objective function and are implemented.

2. A novel integrated redundant control method for path following and yaw stability chassis according to claim 1, characterized by: The first layer optimization distribution strategy in step 2-1, the optimization objective function can be represented by the following formula: wherein denote the 6 optimization control output values of the front-left wheel longitudinal tire force, the front-right wheel longitudinal tire force, the rear-left wheel longitudinal tire force, the rear-right wheel longitudinal tire force, the front wheel steering angle, and the rear wheel steering angle; is a constant matrix, The matrix contains the ideal total tire forces of the upper layer path tracking controller and the ideal yaw moment , q11 is a weight coefficient of the ideal total longitudinal tire force, q12 is a weight coefficient of the ideal total lateral tire force, q13 is a weight coefficient of the ideal yaw moment, respectively, corresponding to the front-left wheel longitudinal tire force, the front-right wheel longitudinal tire force, the rear-left wheel longitudinal tire force, the rear-right wheel longitudinal tire force, the front wheel steering angle, and the rear wheel steering angle. The optimization objective function also needs to meet the following friction limit boundary condition: wherein are the vertical loads of each wheel, respectively, are the side slip angles of each wheel, respectively, in addition to the steering angle of each wheel, which should also satisfy the steering angle physical limit constraints, written in the form of a quadratic programming based on quadratic constraints, which can improve the computational efficiency of the optimization algorithm.

3. The novel integrated redundant control method for path following and yaw stability chassis according to claim 1, characterized in that: The second layer control distribution strategy in step 2-2, the ideal longitudinal tire force of each wheel obtained by the first layer optimization is specifically distributed to the torque of each wheel independent drive motor and brake system, in order to optimize energy loss, when distributing the longitudinal tire force of each wheel, the use of independent drive motor reverse to generate brake torque is preferred, and when the motor reverse torque is insufficient, the brake torque generated by the hydraulic independent brake is used to compensate.

Images