融合预瞄与势场栅格法的紧急避撞驾驶人模型

紧急避障工况下的驾驶人操作具有响应快且动作幅值较大的特点，传统预瞄驾驶人模型已不能适应紧急避障工况的需求，故考虑实际避撞场景开发相应的驾驶人模型就显得尤为必要。针对此种状况，基于驾驶模拟器，结合紧急避撞工况实际驾驶人操纵数据，提出了一种融合预瞄与势场栅格法的紧急避撞驾驶人模型。首先针对紧急避撞工况下车辆运动特点，建立车辆横、纵向耦合非线性动力学模型，并给出其状态空间方程描述；其次，离线仿真分析紧急避撞系统特征，并结合线性二次型最优控制，建立最优曲率预瞄+跟踪误差反馈驾驶人模型；再者，基于紧急避撞工况下真实驾驶人经验转向行为数据，开发基于势场栅格法的驾驶人模型，为进一步提高驾驶人模型对避障行驶工况的适应性，将基于势场栅格法的驾驶人模型与最优曲率预瞄+跟踪误差反馈驾驶人模型进行融合，并基于Sigmoid函数实现两者输出的权重分配；最后，针对所提出的融合预瞄与势场栅格法的驾驶人模型，开展基于避撞台架的驾驶人在环仿真试验以及实车试验。研究结果表明：在紧急避撞工况下，对比最优曲率预瞄+跟踪误差反馈驾驶人模型，融合预瞄与势场栅格法的驾驶人模型输出的转向动作与实际驾驶人行为较为接近，可在保证避障安全性的前提下，兼顾避障路径跟踪精度与车辆行驶的稳定性。

Emergency Collision Avoidance Driver Model Based on Fusion of Preview Model and Potential Field Grid Method

Under the condition of emergency obstacle avoidance, the driver's operation has the characteristics of quick response and large action amplitude; hence, the traditional preview driver model can no longer meet the demands of emergency obstacle avoidance. Therefore, it is necessary to develop the corresponding driver model by considering actual collision avoidance scenarios. Using driving simulator and actual driver control data on emergency collision conditions, an emergency collision avoidance driver model from the fusion of the preview model and potential field grid method was proposed. Firstly, based on vehicle motion characteristics during emergency collision avoidance, the vehicle lateral and longitudinal coupling nonlinear dynamic models were established, and the state space equation was given. Secondly, the characteristics of the emergency collision avoidance system were analyzed by offline simulation, and a driver model combining the optimal curvature preview and tracking error feedback was established based on a linear quadratic regulator. In addition, a driver model based on the potential field grid method was developed according to empirical data on the steering behavior of real drivers during emergency collision avoidance. To further improve the adaptability of the driver model to obstacle avoidance driving conditions, the driver model from the fusion of the preview model and potential field grid method was proposed; and based on a sigmoid function, the weight distribution of the two driver-model outputs was realized. Finally, the driver in-the-loop simulation test based on collision avoidance bench test and real vehicle test was carried out. The results show that the output steering action of the improved driver model is more consistent with actual driver behavior. The accuracy of tracking the obstacle avoidance path and vehicle stability were also taken into account to ensure the safety of obstacle avoidance.