Abstract:

Because of the computational complexity in the signature element based local submap matching for hierarchical Simultaneous Localization and Mapping (SLAM) algorithm, a signature vector based matching approach was proposed in this paper. Since there may be a difference between the two trajectories that the robot passed the same subregion at first time and at second time, the approach was improved by combining a landmark localization technique to realize more reliable loop closing calculation. In addition, the constraints equation was discussed for the case that two or more closed loops share a common subpath for obtaining a global optimized map of the environment. Finally, a demonstration case was given to show the application of our approach.

Key words: sub-map matching;simultaneous localization and mapping;landmark;signature vector