Incorporating environmental measurements in navigation

Abstract

Extended missions in unknown regions present a significant navigational challenge for autonomous underwater vehicles (AUV). This paper investigates the long-term performance of a concurrent mapping and localization (CML) algorithm for the scenario of an AUV making observations of point features in the environment with a forward look sonar. Simulation results demonstrate that position estimates with long-term bounded errors of a few meters can be achieved under realistic assumptions about the vehicle, its sensors, and the environment. Potential failure modes of the algorithm, such as divergence and map slip, are discussed. CML technology can provide a significant improvement in the navigational capabilities of AUVs and can enable new missions in unmapped regions without reliance on acoustic beacons or surfacing for GPS resets.