SLAM within indoor loops by using incremental scan registration

This paper presents an efficient solution to fulfill the Simultaneous Localization and Mapping (SLAM) in indoor environment where there are monotonous areas and large loops. The proposed approach works on Split Sparse Points Maps (SSPM) and is mainly based on incremental scan registration by using the Point-to-Line Iterative Closest Point (PLICP) algorithm. To enhance the robustness of scan registration in monotonous environment, an Incident Angle Fused Metric (IAFM) is introduced during the association procedure of scan registration. Furthermore, with the manner of sub maps joining, a trajectory bending based loop closure approach is applied, which can efficiently eliminate the map inconsistence that is generated from error accumulation of scans registration. Experiment is conducted in real environment and the results obtained from different methods are compared and discussed, which verifies the validation of the proposed methods.