Visualization of the outliers detected by the SLAMICP library

During the navigation of a mobile robot in an unconstrained environment, unexpected objects may appear that block the planned path of the robot. Detecting and identifying these obstacles is key to avoiding possible crashes. The SLAMICP library provides efficient computation of the location of the robot while simultaneously identifying outlier points in a LiDAR scan, which enables higher data throughput. This paper evaluates the obstacle detection capabilities offered by the outlier detection feature of the SLAMICP library. Various test obstacles have been used to assess the ability of the library to differentiate elements that are not part of the original mapped area. These test objects include different boxes representing static obstacles and a walking person as a dynamic obstacle. The results show that all test objects were successfully detected and located within the map. The shape of the static obstacles was reconstructed by adding together the location of the outliers across multiple scans. Using the same method with the dynamic obstacle revealed the trajectory of the walking person.