Modeling typical driving patterns in target trajectory decisions for vehicles traversing roads with complex shapes

Abstract

Deciding a target trajectory within the edges of a driveway is the primary task undertaken by a person driving along a road. Therefore, a trajectory decision model is an essential component of a complete driver model. In this paper, we propose a trajectory decision model that is suitable for complicated mountain roads and racing circuits. We first present a computational strategy of “point selection in a preview cross-section,” and establish four objective functions corresponding to four typical driving patterns (direction control patterns/habits) that could be applied to a mountain road. Subsequently, inequality constraints are set based on the practical driving features. While referring to the decision-making processes for determining a target trajectory in the real world, a rolling-horizon algorithm is proposed for trajectory decision simulation for a mountain road of an arbitrarily complicated alignment, which is also of an arbitrary length. Simulation examples for the trajectory decision for two mountain highways and one Formula 1 circuit are presented at the end of the paper. The simulation results show that the simulated trajectories produced by the proposed models conform to our expectations and are consistent with the trajectories observed in actual driving. Moreover, the decision model for the target trajectory, along with a previously developed trajectory tracking model, can be combined into an improved driver model with a more complete structure: that is, a “preview—decision—tracking” direction control model.