Reactive Multiobstacle Avoidance for Underactuated Marine Vehicles

Abstract

Collision avoidance (COLAV) is a critical component in the control of marine vehicles. In dynamic environments, there is often a limited time to react to a collision, and the information picture is unpredictable. To accommodate these characteristics, this article proposes a reactive algorithm for generating the avoidance maneuvers of underactuated marine vehicles moving in multiobstacle environments with incomplete knowledge. The strategy builds on the principle of deconfliction, and it can be combined with a variety of nominal behaviors. The generated deconfliction maneuvers can consist of both speed and course changes. Explicit safety guarantees are derived through an analysis of the resulting closed-loop dynamics. Simulations and experiments showcase the efficacy of the proposed method and support the theoretical results.