ABSTRACT
Enabling self-localization for autonomous ships using marine radar data is of great importance, particularly when satellite signals are unavailable. This requires radar detectable objects with known locations (landmarks) to assist in inference of own-ship location. For unmanned ships, it is often difficult to search and select appropriate radar landmarks (RLs) to calibrate the position of own-ship. Properly taking account of the issues associated with radar echoes from these selected RLs for own-ship self-localization is also non-trivial. In this paper, we consider a harborage area where the global navigation satellite system (GNSS) based localization approach may suffer from reliability issues arising from multipath interference. Instead of using GNSS, location data (latitude and longitude) from anchored ships that are received from automatic identification system (AIS) is chosen to provide the RLs for self-localization. A recursive self-localization algorithm is proposed to estimate the position of own-ship using AIS provided location data of anchored ships. Simulation and real-data experiments demonstrate that the proposed solution can facilitate self-localization in GNSS-restricted environments.
