Risk-Predictive Path Planning in Urban Autonomous Driving: A Geometric Approach to VRU Crossing

This research introduces a method for autonomous driving systems to safely overtake bicycles in urban environments. It identifies the risk of sudden crossing when overtaking a bicycle and proposes a solution to minimize this risk. The method classifies the situation into three conditions related to the drivable space of the road and speed of the bicycle, and it determines the target speed and positions. The proposed algorithm uses onboard sensor information and assumptions of the bicycle’s virtual sudden-crossing motion, enabling real-time calculations in a practical environment. Simulations demonstrate the method’s effectiveness, showing it can generate a natural overtaking speed and lateral gap based on road width and bicycle speed. The proposed method is compatible with waypoint-based path generation methods we have proposed in previous research, making it a promising solution for future autonomous driving systems. Future research will discuss the method’s implementation in automated vehicles, contributing to safer and more efficient autonomous driving systems.