A Sequential Population-Based Search for AUVs

Autonomous Underwater Vehicles (AUVs) rely on sensor measurements to locate a search target. The sensor measurements are prone to different sources of noise. Noise in measuring and estimating the location of an AUV causes localization errors. These errors can influence the AUV abilities for accurate and safe navigation. Furthermore, AUVs utilize other sensors to trace their target. Noise in these sensors can affect AUVs' performance in locating the target. It can influence their abilities for cooperation. The distribution nature and implicit averaging ability of Evolutionary Algorithms (EAs) can help to overcome the problems introduced by the sensors' noise. However, the limitations on the number of the AUVs and communication errors can be obstacles in utilizing EAs. In this paper, a framework for adapting EAs to a single AUV as a search agent is proposed. EAs concepts are deployed within this framework to implement EA-based algorithms. Simulations were carried out to evaluate the performance of EA-based algorithms in locating a Submarine Groundwater Discharge (SGD) using different cooperative scenarios. The simulations show that the proposed search methods can boost the search performance. They also demonstrate a robust behavior against a range of localization and navigation errors.