A new SLAM method based on SVM-AEKF for AUV

Simultaneous localization and mapping (SLAM) problem is an attractive topic in the AUV research community. This paper presents a new SLAM algorithm based on support vector machines(SVM) adaptive Extended Kalman Filter(EKF) for autonomous underwater vehicle(AUV) to reduce the influence of the change of statistical characteristics of the system noise and the observe noise. First establish a feature based map, then use the EKF to create a map. SVM is employed to generate the adaptive factors according to the ratio of the theoretical covariance to its actual covariance of the innovation sequence. Simulation shows that the improved SLAM algorithm reduces the influence of change of statistical characteristics of noise and enhances the navigation accuracy of SLAM system.