Multi-ride Fusion for Rail Digital Map Construction.

A novel technique for building a detailed Digital Map of the rail environment has been implemented and tested. It is based on the fusion of multiple rides acquired by means of on-board GNSS receivers, IMUs and visual sensors from trains in commercial operation. More in detail, an efficient incremental procedure for processing the collected data to reduce the computational complexity and storage requirements has been proposed. We demonstrate that the proposed system is able to obtain accurate track geometry and trackside objects position information while being robust to signal degradations typical of the rail environment.