Unsuitable lane-change maneuvers are one of the most common potential safety risks in traffic. Nevertheless, it is a maneuver that must be executed carefully by both human drivers and self-driving cars. Developing a suitable automated driving algorithm for self-driving cars to address the lane changing problem is not straightforward because the problem can only be stated as a high dimension problem with many variables and parameters. Therefore, safe driving needs to be assured by optimal trajectory generating algorithms as well as high-level behavioral planners, which assess the safety of the intended behavior and discard the infeasible trajectory candidates. In this paper, a hybrid approach to a behavioral planner algorithm with lane changing behavior using the Frenet coordinate system was developed to solve the lane changing maneuver problem on roads that consist of road segments with different curvature values. The controllers that need to generate the actuator commands according to the reference trajectory cannot always precisely follow the trajectories. Hence, the waypoint generation algorithm has to adapt to the controller and vehicle dynamics. The road structures for the driving scenarios were modeled according to the OpenDRIVE format and were implemented in a model-based traffic simulation environment MOBATSim. The proposed approach is evaluated and demonstrated by the simulating driving scenarios in the same environment.
A Hybrid Approach using an Adaptive Waypoint Generator for Lane-changing Maneuver on Curved Roads
Proceedings
2021-05-06
15 pages
Article/Chapter (Book)
Electronic Resource
German
Lane offset maneuver of autonomous vehicles through curved roads
European Patent Office | 2024
|AIAA | 2004
|