New and better method of underwater LBL localization with unsurveyed transponders

Newman et al studied the problem of underwater LBL (long base-line) localization of AUV (autonomous underwater vehicle) with unsurveyed transponders^[2]. Their method has many shortcomings discussed in the full paper. We aim to eliminate these shortcomings as much as possible with a different and we believe better method by using the unscented Kalman filter (UKF) based SLAM (simultaneous localization and mapping) technique. Adopting the stochastic mapping method, we combine the coordinates of the AUV and transponders into one generalized state vector, which is estimated using UKF with measurements of the distances of the AUV to the transponders. This method is a real-time one and is superior to the method of Ref. 2, which solves the nonlinear measurement equation under constraints of the AUV's kinematics. The simulation results show preliminarily that: (1) our new method doses eliminate much of the shortcomings of Ref. 2; (2) the localization error of AUV is bounded and within ten meters, which is much smaller than those obtained by dead-reckoning and original LBL methods; (3) the estimation error of the location of each transponder also converges to a very small value of a few meters.