Obstacle Recognition Using Multiple Sensors in RoboCup MSL

Abstract

In recent years, autonomous mobile robots have been increasing in commercial facilities and urban environments. In the RoboCup Middle Size League, accurate detection of enemy robot positions is crucial for tactical behavior in high-speed multi-robot environments. Previous obstacle recognition using omnidirectional camera suffered from measurement errors depending on observation distance, especially when enemy robots were densely clustered. This study aims to achieve stable obstacle recognition by combining LiDAR with omnidirectional cameras to construct a high-precision spatial model based on probability distributions.