紧急工况下具有自主权限的车辆应急避撞多目标协调控制方法研究

When drivers dealing with emergency collision avoidance conditions, their inevitable limitations will cause reaction lag, error judgment and wrong operation, which is prone to result to traffic accidents. Advanced driver assistant system (ADAS) can to some extent avoid or mitigate collision accidents by improving drivers' perception and decision-making ability. Nowadays, ADAS does not own the highest authority of the emergency avoidance control system, therefore it cannot fundamentally avoid the improper operation caused by those limitations. In this project, ADAS will be further developed with a innovative idea that intelligent control with independent authority should replaced personnel drive in emergency situations. The project research focus will cover the following aspects. Firstly, in view of emergency avoidance, stability control and real-time requirements, a model of vehicle dynamics system with varying parameters and a generalized "Vehicle-Environment" mechanical system will be accurately established. Secondly, state variables related to ego-vehicle stability will be dynamically established, and at the same time the driving intention of the preceding vehicle will be identified, as well as its driving trajectory will be predicted. Thirdly, the AFS (Active Front Steering) and ESP (Electronic Stability Program) controllable domains will be analyzed to study the intervention time and criteria of the AFS and ESP coupled dynamic systems and the coordination mechanism of task allocation. Fourthly, to realize the dual goal of collision avoidance and stability, a multi-object constraint reactive strategy with no path planning will be adopted to make the multi-object coordination control of collision avoidance. Through the research of this project, the function of ADAS can be greatly expanded, which has great significance for the development and improvement of ADAS.

驾驶员由于其自身局限性在处理紧急避撞工况时容易出现反应滞后、错误判断、错误操作而引起交通事故。先进驾驶辅助系统(ADAS)可以提高驾驶员的感知与决策能力，一定程度上避免或是减轻碰撞事故，然而目前ADAS还没有自主权限的应急避撞控制系统，因此不能根本上避免由于驾驶员局限性带来的不当操作。本项目提出在紧急工况时由具有自主权限的智能控制替代人员驾驶。项目研究内容：针对应急避撞、稳定性以及实时性要求，将建立含多变参数的车辆动力学系统与“车-环境”广义机械系统的精确模型；动态预测自车稳定性相关的状态量，同时辨识前车驾驶意图并预测其行驶轨迹；进行AFS和EPS可控域分析，研究AFS和EPS耦合动力学系统的介入时机和准则、任务分配的协调机理；采用无路径规划多约束的反应式策略进行避撞多目标协调控制，实现应急避撞与稳定性的双重目标。通过本项目的研究，可以扩展ADAS的功能，对发展和完善ADAS具有重要意义。

先进驾驶辅助系统(ADAS)可以提高驾驶员的感知与决策能力，一定程度上避免或是减轻碰撞事故。由于在紧急避撞工况下人员驾驶的局限性及不确定性，目前不具备自主权限的应急避撞控制系统，不能根本上避免由于驾驶员局限性带来的不当操作的问题，因此在紧急工况时由具有自主权限的智能控制替代人员驾驶极有必要。本项目针对应急避撞、稳定性以及实时性要求，首先建立了含多变参数的车辆动力学系统与“车-环境”广义机械系统的精确模型；在此基础上，辨识自车稳定性相关的状态量，同时辨识前车驾驶意图并预测其行驶轨迹；进行AFS和EPS可控域分析，建立AFS和EPS耦合动力学系统的介入准则与协调机制；采用反应式策略进行避撞多目标协调控制，实现应急避撞与稳定性的双重目标。具体而言，项目建立了考虑非线性动力学耦合的动力学模型，设计了基于车载传感器及UKF与PSO相融合的车辆状态辨识模型，建立了基于LSTM的车辆轨迹预测模型，构建了基于相平面法的可控制域，完成了两段的反应式应急避撞与稳定性协调控制。研究结果表明，所设计的反应式避撞策略在保证稳定性的前提下提升了车辆的安全性，同时也满足车辆的实时性需求。本项目的研究，对扩展ADAS的功能，对发展和完善ADAS具有重要意义。