State estimation of multiple autonomous underwater vehicles for wide area survey of seafloor

Abstract

In this paper, we propose a state estimation method for navigation by multiple AUVs (Autonomous Underwater Vehicles) for wide area seafloor observation. The key idea is that moving AUVs estimate their states (horizontal position and heading angle) based on AUVs remaining stationary on the seafloor (landmark AUVs). By alternating the landmark role, each AUV navigates over wide area with small position drift. In our previous work, we demonstrated the case of two vehicles. We now expand this method to more than two vehicles. In the extended method, two AUVs alternate between the moving role and the landmark role, and the other AUVs constantly move and estimate only their own states based on the landmark AUV. To analyse the performance of the proposed method, navigation simulation was done. It was observed that the total time to cover the designated area became shorter as the number of the vehicles increased because every vehicle except a landmark vehicle can navigate simultaneously. Positioning errors were also estimated to be smaller than dead reckoning. Errors in position estimation were found to be about 1.1 m in horizontal position and about 0.6 degree in heading angle after navigating for over 120m × 500m by 4 vehicles. It was demonstrated that AUVs could cover the wide area faster and also achieve stable positioning by using the proposed method. Future works includes formulation of the error propagation in the multiple AUVs navigation.