Multisensor fusion and navigation of mobile robots

Abstract

Navigation and motion control of autonomous mobile robots require an on-line, real-time sensory feedback system to guide robots to stay on planned path, to avoid unexpected obstacles, to return to planned path if a detour is necessary, and even to replan the path. This problem is a difficult one, because of incomplete world knowledge and unknown obstacles. A spherical sensory model similar to human perception is presented. Its spherical octree data structure provides an efficient and unified representation scheme for multisensor fusion, navigation, motion control, and spatial reasoning.