Vision-based adaptive dynamic path planning for mobile robots in crosswalks
Abstract
For the problem of real-time navigation and obstacle avoidance of mobile robots in the scenario of a continuously moving pedestrian on a dynamic zebra crossing, this paper proposes a visual-based adaptive dynamic path planning framework. Firstly, the YOLO algorithm is used to detect dynamic pedestrians in the video stream, and a Region of Interest (ROI) focusing and front obstacle filtering mechanism is designed to filter out redundant pedestrians outside the zebra crossing and behind the robot; Secondly, an optimal safety radius adaptive obstacle expansion strategy is introduced to provide physical safety margin for dynamic obstacles; Then, a lateral penalty-enhanced A* algorithm is proposed to generate smooth and flexible obstacle avoidance trajectories, replacing the traditional aggressive straight paths; Finally, an obstacle change-triggered re-planning mechanism is adopted to balance real-time performance and computational cost. Experimental results based on a 17-second dynamic pedestrian video show that the dynamic obstacle avoidance success rate of this method reaches 100%, and it has lower planning delay and shorter path length compared to traditional algorithms. This framework effectively improves the navigation safety and efficiency in complex dynamic zebra crossing environments.