A Novel Scan Matching Method for Mobile Robot Based on Phase Only Matched Filtering

Two-dimensional (2D) laser point cloud matching of relative positioning plays an important role in Simultaneous Localization and Mapping (SLAM), which is the key component of mobile robots. By calculating the offset between the point sets of current profile and the original profile, relative positioning can be determined. With the rapid growth of data volume, faster and more effective scan matching methods should be published. This paper presented a novel scan matching method where 2D rotation was decoupled by transforming point cloud into Hough Domain (HD). The method based on HD and phase only matched filtering (POMF) performed faster matching speed and higher matching accuracy. POMF was adopted to register the overlapped result, and 2D transform was obtained by performing POMF on the x-axis and y-axis, respectively. Compared with the Iterative Closest Point (ICP), our method has 89.2% improvement in error and a 91.6% advance in the time when dealing with static data. Furthermore, the dynamic data processing experiment demonstrates a lower median and better consistent result.