Research On SLAM Algorithm Based On Information Filter For Underwater Vehicles

Autonomous navigation technology is the research focus among autonomousunderwater vehicle (AUV) related research subjects. Autonomous navigation does notrequire the aid of a priori information about the underwater environment, as when thevehicle moves in the water, its onboard sensors perceive the external environment andextract useful information, then form a map of the environment incrementally whilepositioning itself, this is simultaneous localization and mapping (SLAM). Therefore,SLAM is the key to realize the true autonomous navigation for AUV in unknownenvironment.The EKF-SLAM algorithm is the basic solution to SLAM problem among manyknown SLAM algorithms, which can obtain the best solution to SLAM problem. ButEKF-SLAM describes the uncertain information in SLAM by maintaining acovariance matrix; the calculation efficiency for EKF-SLAM is very slow as it isquadratic in the size of map, which limits its application in large-scale environment.Aiming at the significant computational burden of the EKF-SLAM in the largemap and complex environment, this paper addresses Sparse Extended InformationFilter (SEIF) algorithm based on extended information filter (EIF), which is theinformation form of extended kalman filter (EKF). Through sparsification of theinformation matrix, computational complexity is significantly reduced andSEIF-SLAM can be implemented in constant time irrespective of the size of the map.Moreover the algorithm has a better performance on CPU time and memory usagewhen compared with EKF-SLAM. Finally, MATLAB simulation and a sea trial wasconducted in Tuandao Bay, results and analysis prove the feasibility of the algorithmfor the C-Ranger. In one word, SEIF-SLAM has a high application value inlarge-scale and complex environment.