Robust Alternative Estimation of an Environment Map and Robot Egomotion Using an Omnidirectional Vision Sensor.

摘要

Generation of a stationary environmental map is one of important tasks for robot navigation. Under the assumption of a known motion of a robot, environmental maps of a real scene are successfully generated by monitoring azimuth changes in an image. However, it is difficult to observe the exact motion parameters of the robot because of measurement error of the encoder of the robot. Therefore, observational errors in the generated environmental map accumulated in long movements of the robot. Furthermore, the exact stationary map generation is difficult because of ambiguity of correspondence caused by occlusion.In this paper, we propose a method to generate a stationary environmental map and estimate the egomotion of a robot, recursively, by using an omnidirectional image sensor. And the method can find mismatching and bring to right correspondence by evaluating estimation error of each objects location.