预瞄优化神经网络驾驶员模型

PREVIEW OPTIMIZED ARTIFICIAL NEURAL NETWORK DRIVER MODEL

根据预瞄跟随理论并结合真实驾驶员的学习目标、生理限制与不断优化的行为特性,建立了预瞄优化神经网络驾驶员模型。利用2自由度车辆模型进行了汽车在方向控制下的仿真计算。仿真结果表明,预瞄优化神经网络驾驶员模型与实际合格驾驶员行为非常相似,按照该理论建立的驾驶员模型可以应用于驾驶员-汽车闭环系统的研究和自动驾驶等智能车辆的控制。

A preview optimized artificial neural network (POANN) driver model based on the preview-follower theory with considering the driver's target and his physical restriction are studied. Simulation is taken to study the driver's direction control behavior using 2-DOF vehicle models. The simulating results indicate that POANN driver model is a simple controller that represents the regular driver's behavior and can be used to simulate driver-vehicle close loop system and control of intelligent vehicle, such as autonomous driving.