COGNITIVE ANIMATION SIMULATOR FOR AUTOMATED VEHCLES

针对特定车辆的纵向和横向控制策略成为一组车辆认知动画模拟器的基本算法。首先根据状态方程对车辆进行建模，然后基于滑模控制理论重新构造纵向和横向控制器。这些控制器对车辆参数和路况的不确定性具有很强的鲁棒性，其控制性能可与人类专家驾驶员媲美。例如，侧向控制器对在拥挤的积雪道路上快速变道的车辆实施反转向，从而使车辆沿相邻车道的方向角稳定收敛。利用基本的纵向和横向控制器解决了当前车道车辆跟随、相邻车道车辆跟随、自由行驶、车道跟随、车道变换、停车控制等问题，并用算法表示阐明了多目标车辆下的认知决策问题。在三维虚拟现实动画软件上构建了双车道道路电路，从多个角度可视化了车辆的运动，验证了车辆控制和决策模型。未来的课题包括更多的路况变化、交通事故模拟等。

Longitudinal and lateral control policies for a particular vehicle become basic algorithms for cognitive animation simulator of a group of vehicles. A vehicle is first modeled in terms of a state equation, longitudinal and lateral controllers are newly constructed based on a sliding mode control theory. These controllers are robust with respect to uncertainties of vehicle parameters and road conditions, and yield control performance as good as human expert drivers. For instance, a countersteer is implemented by the lateral controller for a vehicle making a rapid lane change on a packedsnow road, thus yielding a stable convergence of vehicle heading angle along the adjacent lane. Problems such as current lane vehicle following, adjacent lane vehicle following, free driving, lane following, lane change, stop control etc., are solved by using the basic longitudinal and lateral controllers, and cognitive decision aspects under multiple target vehicles are clarified in algorithmic representations. A double-lane road circuit is constructed on a 3-dimension virtual reality animation software, vehicle movements are visualized from various viewpoints, and the vehicle control and decision models are justified. Future subjects include more variation of road conditions, simulation of traffic accidents, and so on.