Active safety-oriented trajectory planning and tracking control method for autonomous vehicles

Abstract

This invention relates to a trajectory planning and tracking control method for autonomous vehicles, focusing on active safety. It addresses issues related to vehicle and pedestrian safety during obstacle avoidance in complex scenarios such as slippery roads, emergency obstacle avoidance, and pedestrian evasion at high speeds. The invention incorporates a traffic environment potential field into the objective function of the MPC controller. Constraints are set, and a predictive model is used to predict the vehicle state at time k+i based on the vehicle state at time k and the control quantity at time k. The minimum objective function value is then used as the optimization objective, along with the set constraints, to obtain the control quantity that satisfies the constraints. This invention improves vehicle lateral and longitudinal safety, ensures pedestrian safety, and exhibits good real-time performance, enabling safe obstacle avoidance in various complex scenarios.

Description

Technical Field

[0001] This invention relates to the field of automotive safe driving technology, specifically to a method for trajectory planning and tracking control of autonomous driving vehicles that takes into account active safety. Background Technology

[0002] With the number of cars on the road increasing year by year, road traffic safety issues are becoming increasingly prominent. Since passive safety technologies such as crash beams, bumpers, and pedestrian safety foam only minimize losses after an accident, people are paying more and more attention to active safety technologies for vehicles—that is, vehicles proactively taking measures to reduce safety risks and thus avoid traffic accidents. The main safety risks of driving vehicles include: first, insufficient analysis of dynamically changing traffic environments, such as those with pedestrians and other vehicles with obstacles, leading to collisions with other vehicles; second, on slippery roads in rain or snow, and in sudden situations, high-speed vehicles are prone to rear-end collisions, skidding, rollovers, and lane departures; and third, collisions caused by pedestrians illegally crossing lanes, which vehicles cannot avoid in time. Although various active safety technologies are widely used in vehicles, such as adaptive cruise control (ACC) and automatic emergency braking (AEB) to improve longitudinal driving safety, and electronic stability program (ESP) and emergency steering assist (ESA) to improve lateral stability, these control units are mainly concentrated at the execution level. In the context of the development of autonomous driving, active safety control needs to coordinate upper-level trajectory planning. Moreover, the interaction and mutual constraints between control units used for longitudinal and lateral safety prevent the overall performance of the vehicle from being optimized. Existing cooperative control mainly focuses on a single safety goal in typical scenarios. Therefore, it is of great significance for autonomous vehicles to coordinate multiple lateral and longitudinal safety goals in various complex scenarios during trajectory planning.

[0003] Huang proposed a method based on vehicle kinematics, combining a model predictive controller (MPC) with an artificial potential field. This method incorporates the traffic environment potential field into the objective function, solving for optimal control with multiple objectives and constraints, while simultaneously performing trajectory planning and tracking control, thus achieving lateral-longitudinal coupled control of the vehicle. Xu Yang and Li H, building on Huang's work, employed a vehicle dynamics model, improving high-speed stability while achieving lateral-longitudinal coupled motion control. Snapper, building on Huang's work, designed a Gaussian-like function to describe the traffic environment potential field. Before optimization, a second-order Taylor formula was used to approximate the potential field function, transforming the nonlinear programming problem into a standard quadratic programming problem, shortening the optimization time and improving obstacle avoidance real-time performance.

[0004] The aforementioned methods simultaneously perform obstacle avoidance trajectory planning and tracking control, achieving lateral-longitudinal coupled control of the vehicle and improving its dynamic obstacle avoidance capability. However, they do not consider the safety issues of vehicles and pedestrians when autonomous vehicles avoid obstacles in complex traffic scenarios. Furthermore, the complex potential field function affects the optimization efficiency of MPC, hindering timely obstacle avoidance. Although using the second-order Taylor formula approximation method can improve computational efficiency to some extent, the resulting errors directly affect control accuracy, which is detrimental to the safety of pedestrians and vehicles in complex traffic environments. Summary of the Invention

[0005] This invention addresses the issues of vehicle safety and pedestrian safety when autonomous vehicles are traveling at high speeds in complex scenarios such as slippery roads, emergency obstacle avoidance, and emergency pedestrian avoidance. It provides a trajectory planning and tracking control method for autonomous vehicles that considers active safety.

[0006] A trajectory planning and tracking control method for autonomous vehicles considering active safety is implemented through the following steps:

[0007] Step 1: The vehicle-mounted sensors acquire traffic environment information, establish a traffic environment potential field based on the traffic environment information, and add the traffic environment potential field to the objective function of the MPC controller;

[0008] Step 2: In the MPC controller, set constraints, including constraints on the vehicle's lateral stability index and control variables;

[0009] Step 3: The MPC controller uses the vehicle's dynamic model as the prediction model. The prediction model predicts the vehicle's state at time k+i based on the vehicle's state at time k and the control quantity at time k. Then, it takes the minimum objective function value as the optimization objective and obtains the control quantity that meets the constraints by optimizing the MPC according to the constraints set in Step 2. That is, the vehicle's front wheel steering angle and the desired acceleration. Finally, it controls the vehicle to complete the safe obstacle avoidance.

[0010] The objective function, which takes the minimum value of the objective function as the optimization objective, is expressed by the following formula:

[0011]

[0012] In the formula, Γ1, Γ2, Γ3, Γ4, R, S, and ρ are the weights of each component; k is the current time; and N is the weight of each component. p and N c P represents the control time domain and the prediction time domain, respectively. road Let P be the potential field function of the road. car,j Let P be the obstacle vehicle potential field function. ped,p Let P be the pedestrian potential field function. goal Let v be the potential field function at the target point.x Main vehicle speed, v ref Let Δu be the desired barrier-free driving speed, ε be the control increment, and ε be the speed at which the vehicle can travel without obstacles. q The relaxation factor for each parameter term in the constraint condition.

[0013] The beneficial effects of this invention are as follows: The control method described in this invention is based on a vehicle dynamics model, considers the safety goals of vehicles and pedestrians in designing the traffic environment potential field and MPC, and combines the two. In the MPC, SQP is used for optimization to directly obtain the control quantity, enabling the trajectory planning and tracking control of the autonomous vehicle to proceed simultaneously. It also coordinates multiple side-longitudinal safety goals in various complex scenarios, thus effectively solving the safety problems of vehicles and pedestrians when autonomous vehicles avoid obstacles in complex scenarios. It has the following advantages:

[0014] I. This invention addresses the lateral stability problem of high-speed vehicles in complex scenarios such as slippery roads, emergency obstacle avoidance, and emergency pedestrian avoidance. It incorporates lateral stability constraints such as the vehicle's lateral load transfer rate into the MPC to reduce the risk of vehicle rollover and skidding, thereby improving the vehicle's lateral safety.

[0015] Second, this invention addresses longitudinal safety issues such as rear-end collisions in complex scenarios by designing an obstacle field that considers the vehicle's longitudinal safety distance. This ensures that the autonomous vehicle maintains a safe distance from the vehicle in front during obstacle avoidance, thus guaranteeing the vehicle's longitudinal safety.

[0016] Third, this invention addresses the problem of pedestrian collision avoidance when pedestrians illegally cross the road. It designs the pedestrian potential field considering the safe distance between pedestrians and vehicles, and uses a combination of steering and braking to avoid pedestrians. This ensures that the vehicle maintains a safe distance from the pedestrian in front during obstacle avoidance, thus guaranteeing pedestrian safety.

[0017] Fourth, in order to enable vehicles to avoid obstacles in a timely manner in case of emergencies, this invention uses a gradient-continuous Gaussian function and a quadratic polynomial to establish the traffic environment potential field, which is beneficial to the optimization solution of MPC. Moreover, MPC is optimized using SQP, thereby ensuring control accuracy and reducing the computation time of optimization solution, and improving the real-time performance of vehicle obstacle avoidance.

[0018] Fifth, this invention improves vehicle side-longitudinal safety, ensures pedestrian safety, and has good real-time performance, thereby enabling vehicles to safely avoid obstacles in various complex scenarios. Attached Figure Description

[0019] Figure 1 This is a block diagram illustrating the principle of the autonomous vehicle trajectory planning and tracking control method considering active safety as described in this invention. Detailed Implementation

[0020] Combination Figure 1 This implementation describes an autonomous vehicle trajectory planning and tracking control method that considers active safety. Based on a vehicle dynamics model and an improved traffic environment potential field, this method obtains control quantities that satisfy the constraints through MPC optimization, enabling trajectory decision-making and tracking control to be performed simultaneously. It also coordinates multiple lateral-longitudinal safety objectives in the potential field model and constraints, namely, satisfying safety requirements such as safe distance between vehicles, safe distance between people and vehicles, and lateral stability.

[0021] The specific process of this implementation method is as follows:

[0022] Step 1: The vehicle-mounted sensors acquire traffic environment information, establish a traffic environment potential field based on the traffic environment information, and add the traffic environment potential field to the objective function of the MPC controller;

[0023] The traffic environment information includes road boundaries, lane lines, obstacle vehicle information, and pedestrian information in the surrounding traffic environment. Combining this information and considering the safe distances between vehicles and between people, a traffic environment potential field is established, and the traffic environment potential field is added to the objective function of the MPC controller.

[0024] In this embodiment, the traffic environment potential field considering the safety objective is mainly used to describe the interaction between the master vehicle (i.e., the autonomous vehicle) and its surrounding environment. When there are obstacles, it guides the master vehicle to smoothly avoid obstacles and move towards the target point; when there are no obstacles, it guides the master vehicle to drive along the center line of the lane. In the past, many potential field functions were designed to be relatively complex and had discontinuous gradients, resulting in discontinuous changes in vehicle dynamics and affecting the efficiency of subsequent optimization solutions.

[0025] Using Gaussian and quadratic functions with continuous and concise gradients as potential field functions can effectively describe complex traffic environments and facilitates the planning of smooth obstacle avoidance trajectories and improves the efficiency of subsequent optimization solutions. Furthermore, to achieve the safety goals of avoiding rear-end collisions between the main vehicle and the vehicle in front, and avoiding collisions with pedestrians violating traffic rules, the potential fields for obstacles and pedestrians are designed separately during modeling, considering the safe distances between vehicles and between pedestrians; that is, the domain of the potential field is set according to the safe distance.

[0026] The establishment of the traffic environment potential field includes an obstacle potential field considering safe following distances and a pedestrian potential field considering safe distances between pedestrians. The specific process for establishing the obstacle potential field considering safe following distances is as follows:

[0027] The obstacle potential field, considering safe following distance, should guide the lead vehicle to change lanes during obstacle avoidance to overtake slower obstacle vehicles ahead, while maintaining a safe distance between vehicles. The closer the lead vehicle is to the obstacle vehicle, the larger the value of the potential field function should be. The longitudinal and lateral domains of influence for the obstacle vehicle should be different; laterally, the lead vehicle can be closer to the obstacle vehicle, while longitudinally, the domain range should be set considering the safe following distance between vehicles.

[0028] A two-dimensional Gaussian function describing the obstacle vehicle's potential field, with its longitudinal coordinate X and lateral coordinate Y as variables, indicates that the potential field function is continuous and its gradient is continuous. The level set of this potential field function is elliptical, which better guides the master vehicle to smoothly change lanes because the master vehicle plans its trajectory along the outer contour of the elliptical level set during obstacle avoidance. As the distance between the master vehicle and the obstacle vehicle decreases, the obstacle vehicle's potential field value increases exponentially; at this point, trajectory optimization will attempt to move the master vehicle away from the obstacle vehicle. The longitudinal and lateral domains of the obstacle vehicle can be represented by the convergence factor δ in the potential field function. X and δ Y To adjust. The obstacle vehicle potential field function is as follows:

[0029]

[0030] Where, η car The potential field coefficient δ represents the potential field coefficient of the obstacle vehicle, which determines the maximum value of the potential field. Y X is the lateral convergence factor of the obstacle vehicle's potential field, affecting the lateral domain range of this potential field. car,j These represent the positions of the main vehicle and the obstacle vehicle in the X direction, and the Y and Y directions, respectively. car,j These represent the positions of the main vehicle and the obstacle vehicle in the Y direction, respectively. δ X It is the longitudinal convergence factor of the obstacle vehicle potential field, affecting the longitudinal domain of the potential field, and its value is the distance scaling factor k. car The product of the reference safe following distance D:

[0031] δ X =k car D (2)

[0032] The reference safe following distance D can be expressed as follows:

[0033]

[0034] In the formula, v x Main vehicle speed, a max Main vehicle braking deceleration, v obs For the speed of the obstacle vehicle, a max,obs t1 is the braking deceleration of the obstacle vehicle, t1 is the perception and response time, and d0 is the minimum safe following distance.

[0035] In this embodiment, the pedestrian potential field considering the safe distance between people and vehicles specifically refers to:

[0036] Traditional pedestrian potential field functions use a circular level set. When the longitudinal domain is too small, vehicles cannot decelerate and steer to avoid pedestrians early enough, resulting in a delayed reaction and overly sharp turns that can easily lead to instability. When the lateral domain is too large, it becomes impossible to optimize a reasonable steering and obstacle avoidance trajectory. Using a pedestrian potential field function with an elliptical level set and adjustable longitudinal and lateral domains not only ensures a safe distance between the vehicle and pedestrians but also guides the vehicle to avoid pedestrians in advance. The pedestrian potential field function is as follows:

[0037]

[0038] Where, η ped δ represents the pedestrian potential coefficient, which determines the maximum value of the potential field. py It is the lateral convergence factor of the pedestrian potential field, affecting the lateral domain of the potential field, δ. px It is the longitudinal convergence factor of the pedestrian potential field, affecting the longitudinal domain of the potential field, X and X Ped,p These represent the positions of the vehicle and the pedestrian in the X direction, and the Y and Y directions, respectively. Ped,p These represent the positions of the main vehicle and the pedestrian in the Y direction, respectively.

[0039] δ py Subject to lateral scaling factor k py Lateral safety distance D between pedestrians and vehicles py The influences and the relationships between them can be represented as follows:

[0040] δ py =k py D py (5)

[0041] Among them, the lateral safety distance D py pedestrian radius R p (Using a circular area to represent pedestrian occupancy of the road) Vehicle width W is limited. c Lateral safety margin D for vehicles and pedestrians m Influence.

[0042]

[0043] δ px Subject to longitudinal distance scaling factor k px Longitudinal safety distance D between pedestrians and vehicles px The influences and the relationships between them can be represented as follows:

[0044] δ px =k px D px (7)

[0045] Among them, the longitudinal safety distance Dpx Subject to the speed v of the main vehicle x and the lateral acceleration a of the main vehicle y Critical lateral safety distance D py Influence.

[0046]

[0047] Step 2: In the MPC controller, set constraints, including constraints on the vehicle's lateral stability index and control variables;

[0048] In this embodiment, the MPC controller considering the security objective is specifically as follows:

[0049] Vehicles in complex scenarios often travel at high speeds and may encounter complex situations such as rain, snow, pedestrians, and vehicles with multiple obstacles. The dynamic nonlinear constraints on the control inputs to the vehicle's actuators, the slippage caused by tire-ground friction, and the roll caused by lateral acceleration are more stringent than in ordinary scenarios, placing higher demands on the vehicle's lateral-longitudinal coupled motion control. Furthermore, autonomous vehicles traveling at high speeds require high real-time performance from the controller, needing to complete trajectory planning and tracking control within a sampling period in the face of dynamically changing traffic environments. Previous obstacle avoidance methods have not adequately considered actual conditions and safety, have relatively limited scenario scope, and do not sufficiently address real-time obstacle avoidance requirements.

[0050] Based on the established traffic environment potential field model, constraints are imposed on lateral stability indices such as the lateral load transfer rate and control variables in the MPC (Multi-Process Control) system to achieve the safety goal of preventing vehicle rollover and skidding under various complex scenarios. During the solution process, the nonlinear dynamic model and nonlinear constraints are linearized, and SQP (Simultaneous Quadrature Optimization) is finally selected for optimization, thereby improving the real-time performance of the MPC controller. Since the optimized solution directly obtains the front wheel steering angle and acceleration as control variables, lateral-longitudinal coupled motion control of the vehicle is achieved.

[0051] In this embodiment, the lateral stability constraints of the vehicle are specifically considered as follows:

[0052] The control objective of the MPC controller is to ensure timely obstacle avoidance while maintaining active vehicle safety. This objective is translated into constraints on the vehicle's main states and the controller output. The optimal control quantity is calculated by optimizing the performance index constraints, and a relaxation factor ε is introduced into the inequality constraints to ensure that a feasible solution can be found under the constraints.

[0053] Vehicle lateral stability metrics include: lateral load transfer rate (LTR) and lateral acceleration a. y The inequality constraints for the sideslip angle β and the yaw rate r are as follows:

[0054] LTRmin +ε LTR V LTRmin ≤LTR≤LTR max +ε LTR V LTR max (9)

[0055]

[0056] β min +εγV βmin ≤β≤β max +εβV βmax (11)

[0057] r min +ε r V r min ≤r≤r max +ε r V r max (12)

[0058] In the formula, LTR min ε is the minimum value of the transverse load transfer rate LTR. LTR ε is the relaxation factor for the transverse load transfer rate. LTR V LTRmin LTR is the minimum relaxation term for the cross-load transfer rate. max ε is the maximum value of the transverse load transfer rate (LTR). LTR V LTR max This is the maximum relaxation term for the load transfer rate; it is used to ensure that the controller can obtain a feasible solution.

[0059] a ymin Let a be the lateral acceleration. y The minimum value, a is the relaxation factor for lateral acceleration. y max Let a be the lateral acceleration. y The maximum value; This is the minimum relaxation term for lateral acceleration. This represents the maximum relaxation term for lateral acceleration.

[0060] β min ε is the minimum value of the centroid sideslip angle β. β The relaxation factor for the centroid sideslip angle β, β max ε is the maximum value of the centroid sideslip angle β. β V βmin ε is the minimum relaxation term for the centroid sideslip angle. β V βmax This is the maximum relaxation term for the centroid sideslip angle;

[0061] r min ε is the minimum value of the yaw rate r. rLet r be the relaxation factor for the yaw rate r. max ε is the maximum value of the yaw rate r. r V r min ε is the minimum relaxation term for the yaw rate. r V r max This is the maximum relaxation term for the yaw rate.

[0062] In this embodiment, the constraints on the control quantity are the output quantity η of the prediction model after considering the actual motion limits, the control quantity u, the control increment Δu, and the inequality constraints that must be satisfied as follows:

[0063] η min ≤η≤η max (13)

[0064] u min +ε u V u min ≤u≤u max +ε u V u max (14)

[0065] Δu min +ε Δu V Δu min ≤Δu≤Δu max +ε Δu V Δu max (15)

[0066] In the formula, η min and η max These are the minimum and maximum values ​​of the output quantity η, respectively. min and u max Let ε be the minimum and maximum values ​​of the control variable u, respectively. u ε is the relaxation factor for the control variable u. u V u min and ε u V u max These are the minimum and maximum relaxation terms of the control variable u, respectively.

[0067] Δu min and Δu max Let ε be the minimum and maximum values ​​of the control increment Δu, respectively. Δu To control the incremental Δu relaxation factor, ε Δu V Δu min and ε Δu V Δu max These are the minimum and maximum relaxation terms for the control increment Δu, respectively.

[0068] Step 3: The MPC controller uses the vehicle's dynamic model as a prediction model. Based on the current vehicle state and future control outputs, the prediction model predicts the vehicle state at future moments, aiming to minimize the established objective function value while satisfying constraints. Through optimization, the control quantities—the front wheel angle and the desired acceleration—are obtained, thereby controlling the vehicle to safely avoid obstacles. The constraints should consider not only the vehicle's dynamic constraints but also its lateral stability constraints, namely, constraints on lateral stability indicators such as lateral load transfer rate, lateral acceleration, center of gravity sideslip angle, and yaw rate, thereby reducing the risk of rollover and skidding and improving vehicle safety and stability. The front wheel angle and steering wheel have a linear transmission ratio. The acceleration controller, based on existing research, uses a fuzzy PID algorithm to obtain the throttle opening and braking pressure to achieve acceleration control.

[0069] Therefore, the essence of obstacle avoidance trajectory planning and tracking control method is to take vehicle safety obstacle avoidance and pedestrian collision avoidance as the main control objectives, while taking into account stability constraints such as lateral load transfer rate, and using front wheel rotation angle and longitudinal acceleration as control variables. It is an optimal control problem with multiple objectives and multiple constraints.

[0070] In this embodiment, the objective function is minimized as the optimization objective, and is expressed by the following formula:

[0071]

[0072] The objective function is mainly divided into four parts: the first part includes the first four terms, which are the components of the potential field of the main vehicle traffic environment. The prediction time domain N is calculated by the prediction model in combination with the potential field function. p The potential field value acting on the main vehicle. By incorporating the traffic environment potential field into the objective function, the trajectory with the minimum potential field value is found to obtain the obstacle avoidance trajectory. The obstacle avoidance trajectory planning mainly depends on the potential field, so the corresponding weight should be relatively large. The second part ensures that the main vehicle maintains the desired speed when driving without obstacles. The third part is used to limit the control increment, with the aim of preventing large changes in the control increment and reducing the movement amplitude of the actuator. The fourth part is the relaxation factor term, to ensure that the controller can obtain a feasible solution. In the formula, Γ1, Γ2, Γ3, Γ4, R, S, and ρ are the weights of each sub-term; k is the current time, N is the weight of each sub-term. p and N c P represents the control time domain and the prediction time domain, respectively. road Let P be the potential field function of the road. car,j Let P be the obstacle vehicle potential field function. ped,p Let P be the pedestrian potential field function. goal Let v be the potential field function at the target point. x Main vehicle speed, v ref Let Δu be the desired barrier-free driving speed, ε be the control increment, and ε be the speed at which the vehicle can travel without obstacles. qThe relaxation factor for each parameter term in the constraint condition. (k+i|k) represents the prediction of time k+i from time k, where k after the symbol "|" indicates that the current time is k.

[0073] The optimal control increment Δu at time k is calculated by combining the constraints and solving the problem. * (k), and thus the optimal control quantity is obtained as follows:

[0074] u * (k)=u * (k-1)+Δu * (k) (17)

[0075] In the formula, u * (k) is the control variable at time k, u * (k-1) is the control variable at time k-1.

[0076] In this embodiment, a Gaussian function and a quadratic function with continuous and simple gradients are first used as potential field functions, which are beneficial for planning a smooth obstacle avoidance trajectory. At the same time, in order to achieve the safety goals of avoiding rear-end collisions between the main vehicle and the vehicle in front and avoiding collisions with pedestrians, the obstacle vehicle potential field and pedestrian potential field are designed with safe distances between vehicles and between people.

[0077] Secondly, constraints were imposed on lateral stability indicators and control variables such as lateral load transfer rate in MPC, achieving the safety goal of avoiding vehicle rollover and skidding in complex scenarios.

[0078] Finally, by employing a gradient-continuous potential function, the nonlinear dynamic model and nonlinear constraints are linearized and then optimized using SQP, thereby improving the real-time performance of the MPC controller. Simulation experiments verify that this method has good versatility in different scenarios. High-speed vehicles can meet the collaborative requirements of multiple safety targets in various complex scenarios when avoiding obstacles, thus achieving safe obstacle avoidance for the vehicle.

Claims

1. An active safety-oriented trajectory planning and tracking control method for autonomous driving vehicles, characterized in that: The method is realized by the following steps: Step one, the vehicle-mounted sensor acquires traffic environment information, establishes a traffic environment potential field according to the traffic environment information, and adds the traffic environment potential field to a target function of an MPC controller; Step two, in the MPC controller, a constraint condition is set, and the constraint condition includes constraints on a lateral stability index and a control amount of the vehicle; Step three, the MPC controller takes a dynamics model of the vehicle as a prediction model, the prediction model predicts a vehicle state at k+i time according to a vehicle state at k time and a control amount at k time, then takes a minimum target function value as an optimization target, and obtains a control amount, i.e., a front wheel steering angle and a desired acceleration, that satisfies the constraint condition through MPC optimization solving according to the constraint condition set in step two, so as to finally control the vehicle to complete safe obstacle avoidance; The target function with the minimum target function value as the optimization target is expressed by the following formula: In the formula, Γ1, Γ2, Γ3, Γ4, R, S, ρ are the weights of each item; k is the current time, N p And N c Respectively control time domain and prediction time domain, P road Is the road potential field function, P car,j Is the obstacle vehicle potential field function, P ped,p Is the pedestrian potential field function, P goal Is the target point potential field function, v x Is the host vehicle speed, v ref Is the expected obstacle-free driving speed, Δu is the control increment, ε q Is the relaxation factor of each parameter item in the constraint condition. 2.The active safety considering autonomous driving vehicle trajectory planning and tracking control method according to claim 1, characterized in that: The traffic environment information acquired in step one includes road boundaries, lane lines, obstacle vehicle information, and pedestrian information. 3.The active safety considering autonomous driving vehicle trajectory planning and tracking control method according to claim 1, characterized in that: In step one, the traffic environment potential field includes an obstacle vehicle potential field considering a safe vehicle distance and a pedestrian potential field considering a safe distance between a vehicle and a pedestrian; The obstacle vehicle potential field considering the safe vehicle distance adopts a two-dimensional Gaussian function taking a longitudinal coordinate X and a lateral coordinate Y of the obstacle vehicle as variables as an obstacle vehicle potential field function, and the obstacle vehicle potential field function is expressed by the following formula: where η car is the obstacle vehicle potential field coefficient, δ Y is the lateral convergence factor of the obstacle vehicle potential field, X and X car,j are the positions of the host vehicle and the obstacle vehicle in the X direction, respectively, Y and Y car,j are the positions of the host vehicle and the obstacle vehicle in the Y direction, respectively; δ X is the longitudinal convergence factor of the obstacle vehicle potential field, which has a value of the distance scaling factor k car times the reference safe distance D: The pedestrian potential field considering the safe distance between the vehicle and the pedestrian adopts a pedestrian potential field function with a horizontal set as an ellipse and adjustable longitudinal and lateral scopes, and is expressed by the following formula: wherein η Ped is a pedestrian potential field coefficient, δ py is a lateral convergence factor for the pedestrian potential field, δ px is a longitudinal convergence factor for the pedestrian potential field, X Ped,p is the position of the pedestrian in the X direction, Y Ped,p is the position of the pedestrian in the Y direction.

4. The active safety considering autonomous driving vehicle trajectory planning and tracking control method according to claim 3, characterized in that: The lateral convergence factor δ of the pedestrian potential field py The relationship between the lateral convergence factor k py And the lateral safety distance D of the vehicle py Is expressed by the following formula: δ py = k py D py The longitudinal convergence factor δ of the pedestrian potential field px The relationship between the longitudinal distance expansion factor k px And the longitudinal safety distance D between the pedestrian and the vehicle px Is expressed by the following formula: δ px = k px D px . 5.The active safety considering autonomous driving vehicle trajectory planning and tracking control method according to claim 1, characterized in that: In step two, the vehicle lateral stability constraints include vehicle lateral load transfer ratio LTR, lateral acceleration a y , side slip angle β and yaw rate r; the specific constraint form is as follows: LTR min + ε LTR V LTRmin ≤ LTR≤ LTR max + ε LTR V LTR max β min +ε β V βmin ≤β≤β max +ε β V βmax r min +ε r V r min ≤r≤r max +ε r V r max where LTR min is the minimum value of the lateral transfer rate LTR, ε LTR is the relaxation factor of the lateral transfer rate, ε LTR V LTRmin is the minimum relaxation term of the lateral transfer rate; LTR max is the maximum value of the lateral transfer rate LTR; ε LTR V LTR max is the maximum relaxation term for the lateral load transfer rate LTR; a ymin is the minimum value of the lateral acceleration a y , is the relaxation factor of the lateral acceleration a y max is the maximum value of the lateral acceleration a y ; is the minimum relaxation term of the lateral acceleration a y , is the maximum relaxation term of the lateral acceleration a y ; β min is a minimum value of the side slip angle of the center of mass, β β is a relaxation factor for the side slip angle of the center of mass, β max is a maximum value of the side slip angle of the center of mass, β β V βmin is a minimum relaxation term for the side slip angle of the center of mass, β β V βmax is a maximum relaxation term for the side slip angle of the center of mass, β r min is a minimum value of the yaw rate r, ε r is a relaxation factor of the yaw rate r, r max is a maximum value of the yaw rate r, ε r V r min is a minimum relaxation term of the yaw rate r, ε r V r max is a maximum relaxation term of the yaw rate r. 6.The active safety considering autonomous driving vehicle trajectory planning and tracking control method according to claim 1, characterized in that: In step two, the constraint on the control amount includes an output amount η, a control amount u, and a control increment Δu of the prediction model, and the specific constraint form is as follows: η min ≤ η ≤ η max u min +ε u V u min ≤u≤u max +ε u V u max Δu min +ε Δu V Δu min ≤Δu≤Δu max +ε Δu V Δu max where η min and η max are the minimum and maximum values of the output η, respectively, u min and u max are the minimum and maximum values of the control u, respectively, ε u is a slack factor for the control u, ε u V u min and ε u V u max are the minimum and maximum slack terms for the control u, respectively. Δu min and Δu max are minimum and maximum values of the control increment Δu, respectively, ε Δu is a control increment Δu relaxation factor, ε Δ u V Δu min and ε Δu V Δu max are minimum and maximum relaxation terms of the control increment Δu, respectively.

Images