A Multi-Stage Trajectory Planning System for Autonomous Vehicle in Mixture Scenarios

Abstract

In the field of autonomous driving, trajectory planning has achieved substantial advancements, ensuring safety and efficiency in the majority of scenarios. Nevertheless, the intricacies of mixture scenarios, marked by the interplay of diverse requirements, present enduring challenges. To tackle these intricacies, we introduce a multi-stage trajectory planning system. The system is comprised of three targeted stages: the path planning stage leveraging a Gaussian process to accommodate static obstacles and curvature limitations, the speed planning stage to address dynamic obstacles and inherent uncertainties, and the decision-making stage to balance safety and efficiency. Each stage is designed to manage specific demands, enabling the system to handle the multifaceted challenges in mixture scenarios. To corroborate the efficacy of our proposed system, we conducted a validation against established algorithms and a simulation under high-speed, long-distance road conditions. The encouraging outcomes underline our system's potential for providing robust and adaptive navigation in mixture environments.