Traffic Rules-Based Decentralized Task Allocation in Autonomous Collaborative Ship Collision Avoidance

This paper addresses the decentralized collaborative collision avoidance problem for multi-vessel encounters. In maritime traffic scenarios, autonomous ships must comply with traffic regulations to prevent potential collision with neighbouring vessels. We consider a two-layer framework: the first layer evaluates collision scenarios, while the second leverages this information to decentralized decide the control action. We enhance the first layer by formulating a rules-based task allocation approach that assigns each vessel either to stand-on or giveway priority based on the collision scenario, ensuring compliance with traffic rules. The method's effectiveness is demonstrated through numerical simulations in scenarios with two different traffic rules, showing its versatility. Additionally, the proposed rules-based task allocation frameworks reduce the computational burden of the second layer for the control optimization compared to a negotiation protocol.