Real-time Detection of Dynamic Obstacle Using Laser Radar

Abstract

Dynamic obstacle detection in unknown environments during mapping is a very essential problem for mobile robots. Dynamic obstacles directly affect the precision of the mobile robot map building. An approach to detecting dynamic obstacles using a 2D laser radar in real-time is proposed in this paper. After filtering out noisy data, sensor readings are mapped into the world coordinate system to build grid map and three consecutive grid maps are maintained. By comparing the state of the same grid cell in the three consecutive grid maps and considering the state of the eight-neighbor cell, static obstacles are identified. Dynamic obstacles then are identified by comparing the current grid map to the already identified static obstacles. The approach has been implemented and validated in real mobile robot MORCS-1.The experimental results have demonstrated that mobile robots can effectively identify dynamic obstacles in unknown environments with good real-time performance and high reliability.