Two-step Method Grid Map Merging Based on Corner Points

Abstract

Multi-robot SLAM (Simultaneous Localization and Mapping) can be used in many collaboration-based operations, such as military rescue tasks. Map merging is the most obvious solution to the multi-robot SLAM. However, traditional corner-based map fusion algorithms failed in complex environments. This paper proposes a corner-based two-step method for map merging in complex environments. First, the map is preprocessed, including orthogonal rotation, filtering, and edge reconstruction. The corners are identified by Harris Corner detection algorithm. By constructing a convex quadrilateral based on the corner points, the overlapping area can be selected for the first step. Then, the intersection of the obtained quadrilateral diagonals is used as a new feature point, and a triangle is constructed for matching to obtain the final overlapping area. Finally, the rotation angle and displacement can be obtained by calculating the transformation matrix. The results show that this method can solve the complex grid map merging problem effectively.