Distributed-Perception-Based Simultaneous Localization and Mapping for Mobile Robots

This paper presents a two-level simultaneous localization and mapping (SLAM) method based on distributed perception that allows us to obtain accurate grid maps of large environments.At local map level,a new local map is built based on information from the robot laser sensor and odometry using a Rao-Blackwellized particle filter (RBPF) method once the robot enters a new camera visual field.Meanwhile,we also solve the camera-calibration problem by using a marker attached to the robot which moves in a curve fashion in the camera visual field.The global level is an adjacency graph whose arcs are labeled with the constraints between local maps.To obtain an accurate and globally consistent map,a stochastic gradient descent (SGD) algorithm is employed to optimize the existed adjacency graph.Experimental results illustrate the validity of the presented approach.