SLAM Problem for Autonomous Underwater Vehicle using SVSF Filter

Fethi Demim, Abdelkrim Nemra, Hassen Abdelkadri, Abdelouahab Bazoula, Kahina Louadj, Mustapha Hamerlain

2018 · 20 citations · 14 references

Concepts

Abstract

This paper addresses the problem of simultaneous localization and mapping (SLAM) for Autonomous Underwater Vehicle (AUV) equipped with a Mechanically Scanned Imaging Sonar (MSIS) as principal sensor operating in structured environments such as marine platforms, harbors, or dams. Our contribution has introduced a method based on the Smooth Variable Structure Filter (SVSF) to solve this problem. The SVSF is known to be very robust to modeling errors and uncertainties. This filter merges the information from various sensors (Doppler Velocity Logs (DVL), the MTi motion reference unit (MRU)) and the observations from the feature extraction based line in order to estimate the vehicle's motion and to construct a map in partially structured environments. The SVSF -SLAM algorithm is implemented with a datasets obtained in an abandoned marine experiment. The proposed algorithm is validated and provides an accurate estimation which gives a good robustness compared to the EKF/NH∞-SLAM algorithms.

References

14