CN110539752B - A kind of intelligent vehicle multi-prediction range model prediction trajectory tracking control method and system - Google Patents

Abstract

Translated fromChinese

本发明公开了一种智能汽车多预测范围模型预测轨迹跟踪控制方法及系统，包括信息感知、路径规划、多预测范围模型预测控制器建模以及驱动执行；环境感知实时采集智能汽车的前方道路信息及自身状态，将相关事件实时收集并传递至多预测范围模型预测控制器供其调用；路径规划根据信息感知的数据规划出一条期望路径；所述多预测范围模型预测控制器能够实现紧急事件下的预测控制；所述驱动执行根据多预测范围模型预测控制器输出的前轮转角值驱动操纵车辆的执行机构。本发明通过调用多预测范围模型预测控制器与可能发生事件相对应的预测时域与相应约束集及成本函数，使得无人驾驶车辆提前改变跟踪路径避免紧急情况的发生从而提升了汽车行驶的稳定性。

The invention discloses a multi-prediction range model prediction trajectory tracking control method and system for an intelligent car, including information perception, path planning, multi-prediction range model prediction controller modeling and driving execution; environment perception collects road information ahead of the intelligent car in real time and its own state, collect relevant events in real time and transmit them to the multi-prediction range model prediction controller for its invocation; the path planning plans a desired path according to the information-aware data; the multi-prediction range model prediction controller can realize emergency events. Predictive control; the drive executes the actuating mechanism for driving and manipulating the vehicle according to the front wheel angle value output by the multi-prediction range model prediction controller. The invention makes the unmanned vehicle change the tracking path in advance to avoid the occurrence of emergency by calling the prediction time domain and the corresponding constraint set and the cost function corresponding to the possible events in the multi-prediction range model prediction controller, thereby improving the driving stability of the vehicle sex.

Description

Translated fromChinese

一种智能汽车多预测范围模型预测轨迹跟踪控制方法及系统A kind of intelligent vehicle multi-prediction range model prediction trajectory tracking control method and system

技术领域technical field

本发明属于智能汽车控制领域，特别涉及了一种智能汽车在预测紧急事件下的控制方法。The invention belongs to the field of intelligent automobile control, and particularly relates to a control method of an intelligent automobile under the condition of predicting emergency events.

背景技术Background technique

行驶智能化作为汽车技术变革核心技术在全球范围内深入展开。现有的模型预测控制大多着眼于路径跟踪的精度的提升上，但是在真实路况上行驶时往往在紧急事件发生的时候才做出相应反应使得无人驾驶车辆无法准确的跟踪期望路径，这是因为在高速行驶工况下车轮纵向力趋近于饱和，在紧急事件发生的情况下转向但轮胎无法提供更多的侧向力，所以车辆在应对紧急事件下容易出现失稳现象。As the core technology of automobile technological transformation, driving intelligence has been carried out in depth around the world. Most of the existing model predictive control focuses on improving the accuracy of path tracking, but when driving on real road conditions, it often responds only when an emergency occurs, so that the unmanned vehicle cannot accurately track the desired path. Because the longitudinal force of the wheel tends to saturate under high-speed driving conditions, and in the event of an emergency, the tires cannot provide more lateral force, so the vehicle is prone to instability in response to an emergency.

环境感知结合V2X技术使得汽车和周围环境之间进行无线通讯，从而为判断紧急事件提供了技术支持。面向无人驾驶的终极目标，要求智能汽车控制系统在复杂工况条件下，具有精确、高效、可靠的控制能力，保证车辆转向稳定性、行驶安全。传统单一预测范围的算法不能解决车辆不同紧急事件下的控制需求和控制功能。Environment perception combined with V2X technology enables wireless communication between the car and the surrounding environment, thus providing technical support for judging emergency events. Facing the ultimate goal of unmanned driving, the intelligent vehicle control system is required to have precise, efficient and reliable control capabilities under complex working conditions to ensure vehicle steering stability and driving safety. The traditional single prediction range algorithm cannot solve the control requirements and control functions of vehicles under different emergency events.

发明内容SUMMARY OF THE INVENTION

一种智能汽车在紧急事件下的控制方法，通过在传统模型预测控制单预测范围的基础上添加多种预测范围的方法来预测可能发生的事件，在跟踪期望路径的同时保持对于如突遇冰雪路面或者行人横穿马路等的应急计划而不是在紧急事件发生的情况下在做出反应。从而提高模型预测控制对于可预见但难以确定发生概率事件的稳定性。A control method for an intelligent vehicle under emergency events, which predicts possible events by adding multiple prediction ranges on the basis of the traditional model predictive control single prediction range, and keeps track of the expected path while maintaining the protection against sudden ice and snow. Contingency planning for pavement or pedestrian crossing, etc. rather than responding in the event of an emergency. Thus, the stability of the model predictive control for predictable but difficult to determine the probability of occurrence of events is improved.

传统的模型预测控制基于环境感知模块的道路摩擦系数估计或先验的摩擦系数，而真实路况中道路摩擦系数难以以准确的数学公式表达；路边行人的行为也难以用概率来进行估计与预测。所以采取环境感知模块结合V2X技术来感知交通道路事件大体情况比如在感知到冰雪天气则提供信息给控制器，控制器则预先调用路面可能出现低附着情况的相关控制策略包括相应的预测时域与约束集；或感知路边有行人的情况下预先调用控制器与之相对应的预测时域与约束集来对可能出现行人横穿马路做出预先的调整等等。通过预先调整使得轮胎的横纵向力都控制在极限范围之内，增加控制系统的鲁棒性和车辆的稳定性。以上复杂工况可以单一出现，也可以同时出现一同处理。The traditional model predictive control is based on the estimation of the road friction coefficient of the environment perception module or the a priori friction coefficient, and the road friction coefficient in the real road conditions is difficult to express with an accurate mathematical formula; the behavior of roadside pedestrians is also difficult to estimate and predict with probability . Therefore, the environment perception module combined with V2X technology is used to perceive the general situation of traffic and road events. For example, when snow and ice weather is sensed, information is provided to the controller. Constraint set; or pre-calling the controller's corresponding prediction time domain and constraint set to make pre-adjustments for possible pedestrians crossing the road when there are pedestrians on the roadside. Through pre-adjustment, the lateral and longitudinal forces of the tires are controlled within the limit range, which increases the robustness of the control system and the stability of the vehicle. The above complex working conditions can occur singly or simultaneously.

本发明的有益效果：Beneficial effects of the present invention:

1、本发明提出的多时域模型预测控制的思想，将原本紧急事件发生后采取操作的模式替换成预先通过环境感知和V2X技术模块判断可能发生的事件，调用多预测范围模型预测控制器与可能发生事件相对应的预测时域与相应约束集及成本函数，使得无人驾驶车辆提前改变跟踪路径避免紧急情况的发生从而提升了汽车行驶的稳定性。1. The idea of multi-time-domain model predictive control proposed by the present invention replaces the original mode of taking action after the occurrence of an emergency with a pre-judgment of possible events through environmental perception and V2X technology modules, and calls the multi-prediction range model predictive controller and possible events. The prediction time domain corresponding to the occurrence event and the corresponding constraint set and cost function enable the unmanned vehicle to change the tracking path in advance to avoid the occurrence of emergencies, thereby improving the stability of the vehicle.

2、本发明提出的综合考虑横纵向力的动力学分段仿射模型，有效的提升模型的精度，解决无人驾驶车辆在复杂工况下控制偏差较大的问题。2. The dynamic segmental affine model that comprehensively considers the lateral and longitudinal forces proposed by the present invention effectively improves the accuracy of the model and solves the problem of large control deviation of the unmanned vehicle under complex working conditions.

附图说明Description of drawings

图1是控制模块逻辑框图；Fig. 1 is the logic block diagram of the control module;

图2是多预测范围框图；Figure 2 is a multi-prediction range block diagram;

图3是多预测范围成本函数框图；Figure 3 is a block diagram of a multi-prediction range cost function;

图4环境感知流程框图；Figure 4 is a flow chart of environmental perception;

具体实施方式Detailed ways

下面结合附图对本发明作进一步说明。The present invention will be further described below in conjunction with the accompanying drawings.

如图1所示，本发明系统的组成包括如下：As shown in Figure 1, the composition of the system of the present invention includes the following:

感知模块Perception module

实时采集智能汽车的前方道路信息、横摆角速度γ、车速V_x、以及质心侧偏角β。并结合V2X短距离感知技术将交通道路信息中如冰雪天气，路边行人或者前车骤停等路况相关事件进行实时收集并传递至控制模块供控制器预先判断调用；与其他模块的配合关系如图4所示。Collect the road information ahead of the smart car, the yaw rate γ, the vehicle speed V_x , and the side-slip angle β of the center of mass in real time. Combined with the V2X short-distance perception technology, the traffic road information such as ice and snow weather, roadside pedestrians, or the sudden stop of the preceding vehicle and other road conditions-related events are collected in real time and transmitted to the control module for the controller to pre-judgment and call; the cooperation relationship with other modules is as follows: shown in Figure 4.

路径规划模块path planning module

根据感知模块所传递的数据规划出一条期望路径，此规划路径不包括对于复杂工况的计算；Plan a desired path according to the data transmitted by the sensing module, and this planning path does not include the calculation of complex working conditions;

动力学建模模块Dynamics Modeling Module

为了适应复杂工况下的精度要求，单一的侧向动力学已经无法满足，所以建立横纵向力的分段仿射动力学模型。同时考虑横向力与纵向力，非线性横纵向力模型进行分段仿射化处理，将轮胎侧偏角α_i与轮胎侧向力F_yi的实际关系曲线分成三段线性表示In order to meet the precision requirements under complex working conditions, single lateral dynamics can no longer satisfy, so a piecewise affine dynamics model of lateral and longitudinal forces is established. Considering both the lateral force and the longitudinal force, the nonlinear lateral and longitudinal force model is subjected to piecewise affine processing, and the actual relationship curve between the tire slip angle α_i and the tire lateral force F_yi is divided into three linear segments.

其中，C_i1和C_i2为第i个轮胎侧向力分段仿射表达式的侧偏刚度；f_i1和f_i2为第i个车轮侧向力分段仿射表达式的常数项；α_pi1，α_pi2为分段点；Among them, C_i1 and C_i2 are the cornering stiffness of the ith tire lateral force piecewise affine expression; f_i1 and f_i2 are the constant terms of the ith wheel side force piecewise affine expression; α_pi1 , α_pi2 is the segmentation point;

将轮胎纵向滑移率k_i与轮胎纵向力F_xi的实际关系曲线分成三段线性表示The actual relationship curve between tire longitudinal slip rate k_i and tire longitudinal force F_xi is divided into three linear segments

其中，K_i1和K_i2为第i个轮胎纵向力分段仿射表达式的纵向刚度；g_i1和g_i2为第i个车轮侧向力分段仿射表达式的常数项；k_pi1，k_pi2为分段点；Among them, K_i1 and K_i2 are the longitudinal stiffness of the ith tire longitudinal force piecewise affine expression; g_i1 and g_i2 are the constant terms of the ith wheel lateral force piecewise affine expression; k_pi1 , k_pi2 is the segmentation point;

得到的动力学模型如下：The resulting kinetic model is as follows:

其中

是汽车纵向路径的导数，

是汽车质心到路径距离的侧向误差倒数，κ为道路曲率，Δψ是航向角误差，V_x、V_y、β、r分别为汽车纵向速度、横向速度，质心侧偏角及横摆角速度，F_xf、F_xr、F_yf、F_yr、m、a、b、Iz。F_xf、F_xr分别为前后轮纵向力，F_yf、F_yr分别为前后轮侧向力，m为车身质量，a、b、Iz分别为前后轴距与绕Z轴转动惯量。in

is the derivative of the longitudinal path of the car,

is the reciprocal of the lateral error of the distance from the car mass center to the path, κ is the road curvature, Δψ is the heading angle error, V_x , V_y , β, r are the longitudinal speed, lateral speed, center of mass side slip angle and yaw angular velocity of the car, respectively,_Fxf ,_Fxr ,_Fyf ,_Fyr , m, a, b, Iz. F_xf and F_xr are the longitudinal forces of the front and rear wheels respectively, F_yf and F_yr are the lateral forces of the front and rear wheels respectively, m is the body mass, a, b, and Iz are the front and rear wheelbases and the moment of inertia around the Z axis, respectively.

控制模块control module

在建立此动力学模型后运用多时域模型预测控制方法进行控制。将step3中所建立的动力学模型转换成状态空间形式并运用一阶欧拉法进行模型的离散化得到After the dynamic model is established, the multi-time domain model predictive control method is used for control. Convert the dynamic model established in step3 into state space form and use the first-order Euler method to discretize the model to get

x_k+1＝Ax_k+Bu_k+dx_k+1 =Ax_k +Bu_k +d

其中状态量x由(V_xV_yrΔψe)组成,综合考虑横纵向动力学特性，A、B为状态矩阵，d为仿射项矩阵；Among them, the state quantity x is composed of (V_x V_y rΔψe), considering the transverse and longitudinal dynamic characteristics comprehensively, A and B are the state matrices, and d is the affine term matrix;

如图2所示，其中x_n类似于传统的确定性模型预测控制，而另外x_d1、…x_dn则代表了各种不同的应急计划，对应于不同的约束与成本函数，如图3所示。这样分开的优点是在无紧急事件时只计算x_n所对应的成本函数并计算输出，在紧急事件下只需计算对应的应急计划即可，出现一种紧急事件即用一种，也可在多种紧急事件同时可能发生时一起计算，从而减小计算负担；A_n、A_dn、B_n、B_dn、d_n、d_dn。A_n、A_dn为状态量矩阵，B_n、B_dn为控制量矩阵，d_n、d_dn为仿射项矩阵。As shown in Figure 2, where x_n is similar to the traditional deterministic model predictive control, and the other x_d1 , ... x_dn represent various contingency plans, corresponding to different constraints and cost functions, as shown in Figure 3 Show. The advantage of this separation is that when there is no emergency, only the cost function corresponding to x_n is calculated and the output is calculated. In the event of an emergency, only the corresponding emergency plan needs to be calculated. When an emergency occurs, one can be used. When multiple emergency events may occur at the same time, they are calculated together, thereby reducing the computational burden;_An ,_Adn ,_Bn ,_Bdn ,_dn ,_ddn . An and A_dn are state quantity matrices, B_n and B_dn are control quantity matrices, and d_n and d_dnare affine term matrices.

建立性能指标如下Establish performance metrics as follows

其中N_p、N_c分别为预测与控制时域，Q、R分别为相应的权重矩阵集，ρ、ε分别为权重系数集和松弛因子集,x_k为状态变量，Δu_k为控制增量。where N_p and N_c are the prediction and control time domains, respectively, Q and R are the corresponding weight matrix sets, ρ and ε are the weight coefficient sets and relaxation factor sets, respectively, x_k is the state variable, and Δu_k is the control increment .

系统约束为：The system constraints are:

其中

in

在k＝0处，

分别为k时刻下各预测范围的初始控制量。At k=0,

are the initial control quantities of each prediction range at time k, respectively.

约束集H、G用来对可能发生的紧急事件进行约束。在U₀处依然是追踪原本的期望路径，同时保持对于突发复杂事件的应急操作。H_n与G_n是无突发事件下的约束集，H_n1、G_n1，H_n2、G_n2，……等是对应不同事件下的不同约束集。如Constraint sets H and G are used to constrain possible emergencies. At U₀ , the original desired path is still traced, while maintaining emergency operations for sudden and complex events. H_n and G_n are constraint sets under no emergencies, and H_n1 , G_n1 , H_n2 , G_n2 , etc. are different constraint sets corresponding to different events. like

V_y,max＝V_xα_r,sat+brV_y,max =V_x α_r,sat +br

其中α_r,sat为后轮饱和时的侧偏角，作为H_n的约束集。where α_r,sat is the side slip angle when the rear wheel is saturated, as the constraint set of H_n .

采用下式作为G_n的约束集。The following formula is used as the constraint set for_Gn .

其中

和

是一组横向误差边界，是沿着避障并保留在道路上的关于路径距离的函数；

近似于车辆等效宽度。in

and

is a set of lateral error bounds as a function of path distance along the obstacle avoidance and remaining on the road;

Approximate vehicle equivalent width.

应对不同的事件，对于约束中的值如α_r,sat、

做出调整，如检测到低温冰雪天气预测在拐角处可能有低附着冰雪路面时，调整权重矩阵Q对横向误差和航向误差做出惩罚使得入弯曲率变小更易进入弯道，权重矩阵R则对前轮转角速率做出惩罚，以维持车辆稳定性，使得控制器得出的最终结果有所不同，从而避免失稳情况。In response to different events, for values in constraints such as α_r,sat ,

Make adjustments. For example, when it is detected that the low-temperature ice and snow weather forecast may have low-adherence ice and snow roads at the corners, the weight matrix Q is adjusted to penalize the lateral error and heading error to make it easier to enter the curve with a smaller curvature, and the weight matrix R is Penalizing the angular rate of the front wheels to maintain vehicle stability allows the controller to produce different final results to avoid buckling conditions.

H和G分别对横摆角速度与质心侧偏角及环境约束进行调整使得事件结束之后继续跟踪原本的期望路径。在检测到路边有行人可能横穿马路事件时调整权重矩阵Q对纵向速度、横向误差和航向误差做出惩罚使得车速降低并略偏向远离行人处行驶，R，H和G则和遇冰雪路面类似做相应调整。如果同时检测多种事件则同时计算单个事件下的输出结果并最终加权求和取得最终结果。如果没有发生紧急情况则一直跟踪期望路径。H and G adjust the yaw rate, center of mass slip angle and environmental constraints, respectively, so that the original desired path can be tracked after the event ends. When it is detected that there are pedestrians on the side of the road that may cross the road, the weight matrix Q is adjusted to penalize the longitudinal speed, lateral error and heading error to reduce the speed and drive slightly away from pedestrians. Adjust accordingly. If multiple events are detected at the same time, the output results under a single event are calculated at the same time, and the final weighted sum is obtained to obtain the final result. If no emergency occurs, the desired path is always followed.

执行模块execution module

操纵执行模块根据控制模块输出的前轮转角值驱动操纵执行机构,执行无人驾驶汽车自主转向,并返回车身状态信息给环境感知模块和动力学建模模块从而使无人驾驶汽车跟踪期望轨迹。The steering execution module drives the steering actuator according to the front wheel angle value output by the control module, executes the autonomous steering of the unmanned vehicle, and returns the body state information to the environment perception module and the dynamics modeling module, so that the unmanned vehicle can track the desired trajectory.

上文所列出的一系列的详细说明仅仅是针对本发明的可行性实施方式的具体说明，它们并非用以限制本发明的保护范围，凡未脱离本发明技艺精神所作的等效实施方式或变更均应包含在本发明的保护范围之内。The series of detailed descriptions listed above are only specific descriptions for the feasible embodiments of the present invention, and they are not used to limit the protection scope of the present invention. Changes should all be included within the protection scope of the present invention.

Claims (4)

Translated fromChinese

1.一种智能汽车多预测范围模型预测轨迹跟踪控制方法，其特征在于，包括：1. a multi-prediction range model prediction trajectory tracking control method for an intelligent vehicle, is characterized in that, comprises:信息感知、路径规划、多预测范围模型预测控制器建模、以及驱动执行；Information perception, path planning, multi-prediction range model predictive controller modeling, and drive execution;所述信息感知实时采集智能汽车的前方道路信息、横摆角速度γ、车速V_x、以及质心侧偏角β，将相关事件进行实时收集并传递至多预测范围模型预测控制器供控制器预先判断调用；The information perception collects the forward road information, yaw rate γ, vehicle speed V_x , and center of mass sideslip angle β of the smart car in real time, and collects relevant events in real time and transmits them to the multi-prediction range model prediction controller for the controller to pre-judgment and call ;所述路径规划根据信息感知的数据规划出一条期望路径，此规划路径不包括对于复杂工况的计算；The path planning plans a desired path according to the information-perceived data, and the planned path does not include calculation for complex working conditions;所述多预测范围模型预测控制器能够实现紧急事件下的预测控制；The multi-prediction range model predictive controller can realize predictive control under emergency events;所述驱动执行根据多预测范围模型预测控制器输出的前轮转角值驱动操纵车辆的执行机构；The driving execution is to drive and manipulate the actuator of the vehicle according to the front wheel rotation angle value output by the multi-prediction range model prediction controller;所述多预测范围模型预测控制器建模包括动力学建模；The multi-prediction range model predictive controller modeling includes dynamics modeling;所述动力学建模的依据是横纵向力的分段仿射动力学模型；所述横纵向力的分段仿射动力学模型为：The basis of the dynamic modeling is the piecewise affine dynamics model of the transverse and longitudinal force; the piecewise affine dynamics model of the transverse and longitudinal force is:将轮胎侧偏角与轮胎侧向力的实际关系曲线分成三段线性表示Divide the actual relationship curve between tire slip angle and tire lateral force into three linear representations

其中，C_i1和C_i2为第i个轮胎侧向力分段仿射表达式的侧偏刚度；f_i1和f_i2为第i个车轮侧向力分段仿射表达式的常数项；α_pi1，α_pi2为分段点；Among them, C_i1 and C_i2 are the cornering stiffness of the ith tire lateral force piecewise affine expression; f_i1 and f_i2 are the constant terms of the ith wheel side force piecewise affine expression; α_pi1 , α_pi2 is the segmentation point;将轮胎纵向滑移率与轮胎纵向力的实际关系曲线分成三段线性表示The actual relationship curve between tire longitudinal slip rate and tire longitudinal force is divided into three linear representations

其中，K_i1和K_i2为第i个轮胎纵向力分段仿射表达式的纵向刚度；g_i1和g_i2为第i个车轮侧向力分段仿射表达式的常数项；k_pi1，k_pi2为分段点；Among them, K_i1 and K_i2 are the longitudinal stiffness of the ith tire longitudinal force piecewise affine expression; g_i1 and g_i2 are the constant terms of the ith wheel lateral force piecewise affine expression; k_pi1 , k_pi2 is the segmentation point;所述动力学建模的表达式为：The expression for the kinetic modeling is:

其中

是汽车纵向路径的导数，

是汽车质心到路径距离的侧向误差倒数，κ为道路曲率，Δψ是航向角误差，V_x、V_y、β、r分别为汽车纵、横向速度，质心侧偏角及横摆角速度；in

is the derivative of the longitudinal path of the car,

is the reciprocal of the lateral error of the distance from the car mass center to the path, κ is the road curvature, Δψ is the heading angle error, V_x , V_y , β, r are the longitudinal and lateral speeds of the car, the center of mass slip angle and yaw angular velocity;所述多预测范围模型预测控制器的建模方法包括动力学模型的离散化：The modeling method of the multi-prediction range model predictive controller includes the discretization of the dynamic model:将动力学模型转换成状态空间形式并运用一阶欧拉法进行模型的离散化Convert the dynamics model to state-space form and use the first-order Euler method to discretize the modelx_k+1＝Ax_k+Bu_k+dx_k+1 =Ax_k +Bu_k +d其中状态量x由(V_x V_y r Δψ e)组成，综合考虑横纵向动力学特性，A、B为状态矩阵，d为仿射项矩阵；Among them, the state quantity x is composed of (V_x V_y r Δψ e), considering the transverse and longitudinal dynamic characteristics, A and B are the state matrices, and d is the affine term matrix;

上述符号：F_xf、F_xr分别为前后轮纵向力，F_yf、F_yr分别为前后轮侧向力，m为车身质量，a、b、Iz分别为前后轴距与绕Z轴转动惯量；x_k为状态变量；A_n、A_dn为状态量矩阵，B_n、B_dn为控制量矩阵，d_n、d_dn为仿射项矩阵。The above symbols: F_xf and F_xr are the longitudinal forces of the front and rear wheels respectively, F_yf and F_yr are the lateral forces of the front and rear wheels respectively, m is the body mass, a, b and Iz are the front and rear wheelbases and the moment of inertia around the Z axis, respectively; x_k is a state variable; An and A_dn are state quantity matrices, B_n and B_dn are control quantity matrices, and d_n and d_dnare affine term matrices.2.根据权利要求1所述的一种智能汽车多预测范围模型预测轨迹跟踪控制方法，其特征在于，所述多预测范围模型预测控制器的建模方法还包括：2. The multi-prediction range model prediction trajectory tracking control method for an intelligent vehicle according to claim 1, wherein the modeling method of the multi-prediction range model prediction controller further comprises:建立性能指标：

Establish performance metrics:

其中N_p、N_c分别为预测与控制时域，Q、R分别为相应的权重矩阵集，ρ、ε分别为权重系数集和松弛因子集，x_k为状态变量，Δu_k为控制增量。where N_p and N_c are the prediction and control time domains, respectively, Q and R are the corresponding weight matrix sets, ρ and ε are the weight coefficient sets and relaxation factor sets, respectively, x_k is the state variable, and Δu_k is the control increment .3.根据权利要求2所述的一种智能汽车多预测范围模型预测轨迹跟踪控制方法，其特征在于，对可能发生的紧急事件进行约束的方法：H和G分别对横摆角速度与质心侧偏角及环境约束进行调整使得事件结束之后继续跟踪原本的期望路径；在检测到紧急事件时调整权重矩阵Q对纵向速度、横向误差和航向误差做出惩罚使得车速降低并略偏向远离紧急事件处行驶；如果同时检测多种事件，则同时计算单个事件下的输出结果并最终加权求和取得最终结果，如果没有发生紧急情况则一直跟踪期望路径。3. a kind of intelligent car multi-prediction range model prediction trajectory tracking control method according to claim 2, it is characterized in that, the method for constraining possible emergencies: H and G are respectively the yaw rate and the center of mass side deflection Adjust the angle and environmental constraints to continue to track the original expected path after the event is over; when an emergency event is detected, adjust the weight matrix Q to penalize the longitudinal speed, lateral error and heading error to reduce the vehicle speed and drive slightly away from the emergency event. ; If multiple events are detected at the same time, the output results under a single event are calculated at the same time and the final weighted sum is obtained to obtain the final result. If no emergency occurs, the desired path is always tracked.4.根据权利要求1所述的一种智能汽车多预测范围模型预测轨迹跟踪控制方法，其特征在于，所述驱动执行在执行完自主转向后，还能够返回车身状态信息给环境感知部分和动力学模型使得智能汽车跟踪期望轨迹。4. The multi-prediction range model prediction trajectory tracking control method for an intelligent vehicle according to claim 1, wherein the drive execution can also return vehicle body state information to the environment perception part and the power after the autonomous steering is performed. The learning model enables the smart car to track the desired trajectory.

Images