Title:A Risk and Comfort Optimizing Motion Planning Scheme for Merging Scenarios

Abstract:Motion planning for merging scenarios accounting for measurement and prediction uncertainties is a major challenge on the way to autonomous driving. Classical methods subdivide the motion planning into behavior and trajectory planning, thus narrowing down the solution set. Hence, incomplex merging scenarios, no suitable solution might be found. In this work, we present a planning scheme that solves behavior and trajectory planning together by exploring all possible decision options. A safety strategy is implemented and the risk of violating a safety constraint is minimized as well as the jerk to feature a risk and comfort optimal trajectory. To mitigate the injection of noise into the actual trajectory, a new analytical trajectory generation method is derived and its optimality is proven. The decision capability is evaluated through Monte-Carlo simulation. Furthermore, the calculation time is evaluated showing the real-time capability of our approach.