3D LiDAR SLAM and Map based Illegal Parking Detection for Electric Scooter

Personal mobility rental service has seen a rapid increase in users due to its convenience of use and the advantage of being able to operate anywhere. One of the most well-known of these services is shared electric scooter (e-scooter). However, the e-scooter's convenience often leads to misuse; users frequently leave them in inappropriate places such as driveways and sidewalks, which is definitely considered illegal. This also causes inconvenience for pedestrians or road users, and in the worst case, it can lead to a safety accident. In this paper, we propose a system for a patrol robot that uses 3D LiDAR simultaneous localization and mapping (SLAM) and map-based illegal parking detection for e-scooters. We leverage the 3D LiDAR and SLAM to create a 3D map that includes information on e-scooter parking zones. Consequently, we estimate the initial position through 3D global localization on the generated 3D map. We utilize semantic segmentation to detect e-scooters and verify whether they are parked illegally. We validated our system on the real scenario dataset and achieved a reasonably well result.