The SLAM algorithm based on PNN in the application for autonomous underwater vehicle

In the underwater navigation and position, if AUV only relies on Inertial Navigation System to navigation and position, the error will accumulate over time. With the SLAM algorithm when AUV in position at the same time it constantly create incremental characteristics map aided navigation. Using sonar sensors to collect submarine characteristic information, there are some areas in the map cannot be very good interpretation all of the current environment under the sea. When the collected position information can be joined into map to match, if AUV navigates in these waters for a long time, which requires a good auxiliary navigation system to correct the location error of the INS system. Aiming at the problem of the map can't accurate position , this paper proposes a Probabilistic Neural Network(PNN) based on the Simultaneous Localization And Mapping(SLAM) algorithm. Take position error information as the input samples to train, through the network nonlinear transfer function , the output information to correct system error. Training data shown that the probabilistic neural network based on SLAM algorithm is used for the feasibility of AUV positioning, at the same time also proved the algorithm's robustness and superiority.