A Configurable Greedy Planner for Collaborative Robotic Search in Acoustic Fish Telemetry Surveys
Abstract
This paper provides a detailed description of the target search problem faced by a system of robotic surface vehicles designed for fish telemetry surveys, specifically in their efforts to establish initial contact with acoustically tagged individuals residing in unknown locations within a freely defined sea area. Based on the discrete prior target distribution and the search cost function of the robotic searcher, a greedy search planner is developed to generate search plans that balance the expected benefits of searching in areas with a higher likelihood of target occupancy with the cost of reaching those areas. A novel approach built on spatial operators contained in a spatial database is presented, enabling an algorithm implementation that extensively utilises spatial predicates such as intersecting and touching geometries. The approach facilitates the use of readily available information from electronic navigational charts to exclude unnavigable areas from the search and enables the generation of any regular tessellation that covers the search area. It also makes the search algorithm independent of cell shape and size, providing flexibility for the operator of the robotic system. Through a range of example search problems set up for realistic marine environments, the algorithm is shown to successfully generate feasible and efficient full coverage collision-free search plans for autonomous fish telemetry surveys.