A particle filter approach for AUV localization

This paper presents a sonar-based localization approach for an autonomous underwater vehicle, in structured and unstructured environments. The system is based on a particle filter approach to represent the vehicle state and it uses a mechanically scanned profiling sonar, acquiring range profiles. A modification to the standard particle filter algorithm is proposed, in order to explore the state space in a more effective way and to reduce computational complexity. The proposed system was validated both in simulation and in trials involving a real vehicle, showing a high robustness and real-time capabilities.