Map building using helmet-mounted LiDAR for micro-mobility

This paper presents a point-cloud mapping method using a light detection and ranging (LiDAR) mounted on a helmet worn by a rider of micro-mobility. The distortion in LiDAR measurements, which is caused by motion and shaking of micro-mobility and rider, is corrected by estimating the pose (3D positions and attitude angles) of the helmet based on the information from normal distributions transform-based simultaneous localization and mapping (NDT SLAM) and an inertial measurement unit. A Kalman filter-based algorithm for the distortion correction is presented under the assumption that the helmet moves at nearly constant translational and angular velocities in any directions. The distortion-corrected LiDAR measurements are mapped onto an elevation map, and the measurements relating to stationary objects in the environments are extracted using the occupancy grid method. The stationary object measurements are utilized to build a point-cloud map. The experimental results in a campus road environment demonstrate the effectiveness of the proposed method.