Obstacle detection and avoidance for autonomous bicycles
Mingguo Zhao, Sotirios Stasinopoulos, Yongchao Yu · 2017
Obstacle detection and avoidance is an essential part in the field of autonomous vehicles. In this paper, we propose a novel methodology for obstacle detection and avoidance for the rapidly-developing category of autonomous bicycles. This comprises the bicycle-specific design of our Light Detection And Ranging(LiDAR)-based sensor system and our obstacle detection technique, including the segmentation of the surrounding environment 3D point cloud into discrete obstacle clusters within the bicycle's path and a method for detecting areas containing liquids on the road surface. Moreover, our obstacle avoidance technique is introduced, consisting of a local path planning and a path following part, enabling our bicycle to avoid the closest obstacle lying on the bicycle's user-defined route. The proposed method takes into account the unique dynamic characteristics of bicycles, and the results of the simulation and outdoor experiment confirm its efficacy and promising behavior.