A market-based task allocation framework for autonomous underwater surveillance networks

Abstract

Realisation of underwater robotic surveillance networks raises several challenges for marine robotics. The underwater scenario is typically characterised by intermittent and unreliable communication. This makes it challenging to develop task allocation schemes suited to work effectively in underwater surveillance applications. We propose a market-based approach to task allocation, which works in a completely distributed way. Through periodic auctions, the algorithm achieves the dynamic assignment of robots to tasks throughout the mission. There is no central auctioneer and any robot becomes an auctioneer when it intends to execute a task. Through periodic auctions, all the robots are sequentially allocated to the tasks. The algorithm is designed to increase the robustness to poor communication and to allow task reallocation, to adapt the allocation to the evolving scenario. Results from computer simulations are reported that support the proposed approach. An Anti-Submarine Warfare application is considered to test the scheme. In this application, the surveillance of areas of different dimensions has to be accomplished by a team of AUVs.