An AUV Cooperative Target Localisation Strategy with Bearing-Only Measurements Based on Bayesian Occupancy Grid Mapping

We propose a cooperative adaptive behaviour to control multiple underwater robots for localising and tracking targets using bearing-only measurements. The behaviour uses as perception layer an Occupancy Grid (OG) Mapping-based framework presented in our recent work. The produced maps show the probability of target presence at different locations. This information is exchanged and fused between the robots to produce maps that allow to estimate the x - y target position. Using these OG maps, the robots make non-myopic coordinated decisions for their heading angles to create favourable geometric network configurations. The reached configurations increase target probability of detection and improve target localisation. The developed control framework is generic, distributed in nature and is suited to control the underwater vehicles of the passive sonar network under development at CMRE. We report results of nontrivial simulations of the developed method that demonstrate its effectiveness in controlling two underwater robots equipped with passive sonars in a realistic underwater surveillance scenario.