Sensor based navigation and localization methods for autonomous underwater vehicles [electronic resource]

Abstract

An algorithm designed to navigate an Autonomous Underwater Vehicle (AUV) within a charted environment is presented. The algorithm processes sensor inputs from the AUV high resolution scanning sonar, compass and velocimeter. The operating environment is modeled with a suitable three dimensional potential function and its gradient which form an attractive field. This algorithm provides performance comparable to the Kalman Filter with the advantage of reduced computational requirements facilitated by precomputation and table look up of correction factors. Applications of data smoothing filters and sliding mode theory are also investigated. The applicability and robustness of this approach are demonstrated with actual test data obtained with the NPS Phoenix submersible and extended simulation of complex environments including unmodeled obstacles.