Merging grid maps via point set registration

This paper addresses the issue of merging 2D grid maps via point set registration. It first turns the map merging into the problem of registering point sets with outliers including noises and missing data. Then, it presents the corresponding objective function by introducing an overlapping percentage for partial registration, which can be solved by the proposed iterative closest point (ICP) algorithm. This algorithm can automatically compute the merging parameters with the overlapping percentage, and it has been proven to converge monotonically to a local minimum from any given initial parameters. To get the global minimum, good initial parameters are required, which are successfully estimated in this paper. Furthermore, we discuss and present the computational complexity of the proposed ICP. Experimental results carried out with real robot data sets demonstrate the robustness and accuracy of our approach over previous methods.