Robotic Mapping and Detection of Dynamic Objects in Outdoor Environments

This paper presents an approach for robotic mapping in dynamic environments. Thanks to a detection method based on the notion of free space, scanned points are classified as belonging to static or moving objects. The detected dynamic points are filtered from the point cloud of the environment, and a 2D heat map showing the distribution of the moving objects in the surveyed area is built. Experimental tests are carried out with an autonomous tracked mobile robot equipped with a 3D LiDAR sensor, a 9-axis IMU, and a GNSS receiver in a parking lot. The results demonstrate the feasibility of the approach in a mapping task with dynamic objects.