An Occupancy Grid Map Merging Algorithm Invariant to Scale, Rotation and Translation

Abstract

In this paper, we consider the problem of merging occupancy grid maps obtained by different autonomous robotic agents exploring the same environment. These robots may have different sensors, processing power, memory capacities, and mapping features. Maps produced can present variations on scale, accuracy or orientation. Nowadays, merging such maps is a challenge which led us to propose an alternative heuristic approach which considers these variations. The algorithm implemented uses Scale Invariant Feature Transform (SIFT) to detect key-points while calculating transformations (rotation, translation, and scale) to merge the maps. Merging is accomplished without a priori information about a robot's initial position and orientation. Public available data sets were used to test the algorithm, and it produced reliable combined maps. Finally, an analysis based on different tests is presented and discussed.