Optimal preview trajectory decision model of lane-keeping system with driver behavior simulation and Artificial Potential Field

Almost all lane-keeping system directly use the lane central line as the preview trajectory to control vehicle tracking it, this simple method has no coherent to driver behavior because real drivers usually don't keep vehicle strictly tracking lane central line but always make corner-cutting during turning. In this paper, from the point of view of the simulation of the real driver behavior, especially driver preview characteristic, and with Artificial Potential Field method, a new preview trajectory decision approach is proposed, in which a total energy function of artificial potential field with two evaluation indexes of safety and maneuverability is established to dynamically deciding vehicle preview trajectory with lane marker, vehicle current state and vehicle steering limit. The simulation of different lane shape such as circle lane, snake lane and double-lane-change and low and high velocity show that this model can dynamically decide the optimal trajectory with lane width and curvature coming from lane marker information and has good occupant acceptance.