Results from the simulation of a decoupled concurrent mapping and localization technique

Increasing interest by the United States Navy in non-traditional navigation algorithms for unmanned undersea vehicles (UUV) has lead to the development of decoupled concurrent mapping and localization (DCML) techniques. DMCL, though having unbounded errors, provides a potential navigational aid that can increase covertness and limit the exposure of an unmanned undersea vehicle. This paper presents the simulation results of a DCML technique developed by the Navy. The first portion of the paper summaries the DCML algorithm original published in 1998. This technique utilizes a Kalman filter to estimate the location of the vehicle based upon features identified from a forward-looking sonar. A comprehensive vehicle simulator is used to generate the simulated sonar returns as range and bearing to targets. The feature characterization and association process groups and identifies common features between each ping. Lastly, a MATLAB simulation of the DCML technique is presented. Several results are presented from short straights run to longer runs that follow a "mower" type pattern over several thousand feet.