基于人-车-路-环境识别与信息交互的四轮独立驱动电动汽车纵向避碰预警与协同控制

As the four-wheel-independent-drive electric vehicle (4WID EV) will run with other kinds of vehicles in the future transportation system, its safety may be one of the key problems before massive application. Therefore, this research proposal plans to analysis the coupling relationships and interaction mechanisms of the driver-vehicle-road-environment system. The results are utilized to construct a 4WID EV in-the-loop driving simulator system, which enables multiple drivers to attend in the same scenario through network. The experimental data are adopted to explore the association between the longitudinal driving behavior characteristics and traffic safety. The driver mental state is identified by multi-modal information fusion method to develop the safe distance model and the hazard situation assessment model based on the results of environment identification and driving behavior. Once there exists longitudinal collision possibility, a multi-objective longitudinal cooperative control method for 4WID EV is studied based on dynamic control allocation and multi-objective genetic algorithm, whose controlling objectives include the safety, the economy, the comfort and the traffic density by considering the redundancy of the motors and the multi-domain coupling phenomena between mechanical and electromechanical elements. The research achievements can improve the accident prevention and collision avoidance capability of 4WID EV and make full use of its energy efficiency and environmental protection advantage, which provides technical support and theoretical reference to the development of safety, comfort, energy saving and environmental friendly intelligent electric vehicle.

针对四轮独立驱动电动汽车在未来道路交通系统应用中可能存在的安全问题，本项目通过分析人-车-路-环境相互耦合关系和作用机理，构建多驾驶员联网的四轮独立驱动电动汽车在环模拟驾驶系统并对实验数据进行统计分析，探索驾驶员纵向驾驶行为特性与行车安全之间的关联模型，采用多模信息融合方法对驾驶员精神状态进行辨识，开发基于环境识别与驾驶行为的安全距离和危险态势评估模型，当存在纵向碰撞危险时，以行驶安全性、车辆经济性、乘坐舒适性和交通顺畅性为控制目标，同时考虑其执行器冗余和机电多域耦合现象，基于动态控制分配和多目标遗传算法研究能适应不同驾驶员特性的四轮独立驱动电动汽车纵向协同控制方法，在发挥其节能环保优势的同时提高其事故预防与主动避障能力，为研发安全、舒适、节能与环保的智能型电动汽车提供技术支撑和理论参考。

针对四轮独立驱动电动汽车在未来道路交通系统应用中可能存在的安全问题，通过分析人-车-路-环境相互耦合关系和作用机理，探索驾驶员纵向驾驶行为特性与行车安全之间的关联模型，开发基于环境识别与驾驶行为的安全距离和危险态势评估模型，研究能适应不同驾驶员特性的四轮独立驱动电动汽车纵向协同控制方法。主要研究成果如下：（1）开发并完善了车载信息采集与环境感知系统、驾驶模拟系统，基于模式识别理论研究了环境信息检测与识别算法，开展了车道线识别、车辆和行人等障碍物检测算法研究，为电动汽车更好地理解所处的交通环境提供有用等信息；（2）建立多驾驶员联网的四轮独立驱动电动汽车在环模拟驾驶系统，采集的驾驶员视觉和生理信号，建立基于隐马尔科夫模型的驾驶行为与意图辨识模型，基于驾驶安全性形成机理，基于BP神经网络理论构建基于能力匹配的驾驶疲劳险态辨识模型；（3）根据环境感知和驾驶员行为分析结果，建立了纵向运动状态下驾驶员的行为动力学模型，得到人-车-路-环境的协同作用机理，提出一种基于二阶TTC的自动紧急制动与避碰模型；（4）联合使用CarSim与Simulink搭建了驾驶员闭环控制的四轮独立驱动电动汽车仿真模型，开发一台四轮毂电机驱动的原理型样车，用于验证所开发的算法；（5）在软约束模型预测控制的框架下兼顾安全性、乘坐舒适性和经济性提出多目标约束柔化的自适应巡航ACC控制策略，以满足自适应巡航车辆的行驶需求；（6）研究了四轮独立驱动电动汽车在线性稳定工况、非线性紧急工况、转向工况及基于电机Map图下的力矩分配算法；（7）提出了无人驾驶四轮独立驱动电动汽车轨迹跟踪分层控制策略，提高其在高速工况和低附着工况的行驶稳定性。本项目研究成果在发挥其节能环保优势的同时提高其事故预防与主动避障能力，为研发安全、舒适、节能与环保的智能型电动汽车提供技术支撑和理论参考。