BUILDING, REGISTRATING, AND FUSING NOISY VISUAL MAPS

The problem of building three-dimensional descriptions (visual maps) of the environment of a mobile robot using passive vision is considered. Since noise is present, they incorporate information about the geometry of the environment and about the uncertainty of the parameters defining the geometry. This geometric uncertainty is directly related to sensor uncertainty. We show how visual maps corresponding to different positions of the robot can be registered to compute a better estimate of its displacement between the various viewpoint positions, assuming an otherwise static environment. We use these estimates to fuse the different visual maps and reduce locally the uncertainty of the geometry primitives which have found correspondents in other maps. We propose to perform these three tasks (building, registrating, and fusing visual maps) within the general framework of extended Kalman filtering, which allows efficient combination of measurements in the presence of noise.