A Global Localization System for Mobile Robot Using LIDAR Sensor

Abstract

This paper demonstrates a set of approaches addressing mobile robot localization problems. The approaches can operate efficiently in 2D occupancy grid map created by robot which is equipped with LIDAR sensor. Given a complex environment and a known map, the robot could be placed at anywhere initially and the global localization system described in this paper can be applied to startup stage in which robot can realize its initial location and orientation in the map without any movement. In the following stage, the system can track the moving robot's position in real time. Thus as long as the robot start the navigation process, the robot system knows where the robot is at the beginning and the robot can be navigated to a reachable place directly. Combined with other sensors, such as cliff sensor, the robot can perceive the kidnapped problem, and then relocate the position to recover from global localization failures. The experimental results show that compared with appointing the initial position and orientation manually, it costs seconds in startup stage to figure out the robot's initial pose in the map, and the system can update the robot's position in motion in real time.