Detection of micro-mobility vehicle with thin frame by human-vehicle integrated deep-learning model

One common problem of light-detection-and-ranging (LIDAR) sensors is that the data acquired has sparse parts. Therefore, when trying to detect micro-mobility with thin frames from such point cloud data, it becomes difficult due to a lack of the amount of point clouds that indicate micro-mobility. This paper proposes a system that can detect micro-mobility vehicles with thin frames from the data acquired by multiple LIDAR sensors by treating the micro-mobility and the person riding it as a single object. In order to demonstrate this, we acquire data that can grasp the state of micro-mobility using multiple LIDAR sensors, perform machine learning on the acquired data, and evaluate the detection accuracy.