A-NetShoot: Adaptive vessel guidance for purse seine net shooting

Purse seining captures large pelagic fish schools by shooting a net from a vessel that unfurls vertically to form a cylindrical wall around the fish. During the shooting phase, precise vessel guidance is crucial, especially against unpredictable currents and evasive, free-swimming fish. These factors are difficult to predict and are traditionally left to the discretion of experienced captains. As the industry moves toward intelligent systems, explicit planning is required to improve success rates and reduce operational costs. This paper frames net shooting as the deployment of an underactuated, length-increasing deformable linear object and introduces A-NetShoot, an optimization-based adaptive path planning framework for vessel guidance. A-NetShoot operates in two steps: first, it estimates the evolving net shape by tracking sparse buoy-mounted sensors and solving for the discretized net geometry via an online least-squares problem. In parallel, sonar observations are modeled with a Gaussian mixture to estimate fish school location and distribution. Using these estimates, the algorithm forecasts net drift and fish motion over time. Then, it refines the vessel trajectory to maximize enclosure success. Extensive simulations under varied currents and fish behaviors show that A-NetShoot outperforms state-of-the-art methods, demonstrating the potential of adaptive guidance based on net-geometry feedback.