Accurate and reliable trajectory prediction is essential for autonomous vehicle (AV) navigation in complex traffic scenarios. However, traditional trajectory prediction methods may fail in dynamic and unpredictable environments, resulting in poor planning and increased safety risks. To address these challenges, this paper presents Dynamic Occupancy Set-Aware Planning (DOSP), a novel framework that integrates Dynamic Occupancy Sets (DOS) with motion planning. DOSP enhances future space estimation through DOS extension, accounting for obstacle orientation and shape to predict more accurate occupancy regions. By integrating DOS predictions into both the constraints and cost functions of the planning process and employing a multi-time domain planning configuration, DOSP ensures proactive hazard avoidance and safer navigation. Extensive evaluations on the CommonRoad platform demonstrate that DOSP significantly improves success rates and reduces collision rates compared to traditional methods. These findings underscore DOSP's potential as a robust solution for AV decision-making in uncertain and dynamic environments.
Dynamic Occupancy Set-Aware Planning for Autonomous Vehicles
24.09.2024
2553559 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch