Multi-objective trajectory design for overtaking maneuvers of automated vehicles
Tamás Hegedüs, Balázs Németh, Péter Gáspár · IFAC-PapersOnLine · 2020
The paper proposes a trajectory design method for overtaking maneuvers, in which several performances are incorporated. Thus, it leads to a multi-objective optimization task. It is based on a simple mathematical representation of feasible trajectories by using a potential field approach. The solution of the complex computations in the optimization is approximated by neural networks. The proposed solution is based on the idea that there are a large number of collision free feasible trajectories. The goal of the method is to find the trajectory, which is suitable for the specified objectives. The result of the paper is an integrated decision for both longitudinal and lateral directions, such as longitudinal acceleration command and the values of the defined path. The effectiveness of the method is presented through CarMaker simulation environment.