Method for obstacle detection and map reconfiguration in wheeled mobile robotics

This paper describes a method for obstacle detection, map building and map reconfiguration for autonomous mobile robotic application, this proposal is based on a stereoscopic vision system and epipolar geometry to achieve obstacle detection, map building and map reconfiguration is presented. A tridimensional projection of every scene in front of the mobile robot is estimated based on geometrical relationship between the environment world coordinates (x,y,z) and their projections through the 2D images captured by the on board stereoscopic vision system of the mobile robot. Depth measure for obstacle distance estimation is calculated by epipolar geometry. The proposal for implementation stereoscopic photometric correspondence problem, is solved in an efficient way with the implementation of the adaptive window for candidate matching (AWCM) sensor, motion control and algorithms are proposed to be implemented in FPGA technology to obtain high performance overall processing.