Case Study on Model Free Determination of Optimal Trajectories in Highly Automated Driving

摘要

Developments regarding driver assistant systems, in the scope of highly automated driving, has gained significant attention during the last years. As a result, planning the movement of the vehicle considering all road conditions and driver/passenger comfort became a crucial issue in that sense. In this paper, a trajectory planning concept that is based on the analytical solution of a given optimization problem is presented where the cost function includes jerk and acceleration of the longitudinal and lateral movements in order to increase the comfort of the passengers. The error of the output trajectory is evaluated over a reference curve that is determined using polynomial fitting approaches. Since the structure of the optimal trajectory is fixed, it is only required to determine the coefficient terms. Therefore, the presented approach is also beneficial from the computational complexity point of view. Real-time experiments have been carried out with our test vehicle on realistic scenarios. The experimental results demonstrated the capabilities and effectiveness of the proposed trajectory planning framework.