Reinforcement Learning for Fishing Route Planning and Optimization
Sammendrag
Pelagic fishing is one of the primary economic activities in the Arctic, especially in the Barents Sea region. Optimizing fishing vessel voyage planning is a feasible avenue for enhancing efficiency, profitability, and sustainability while mitigating accidental risks posed by adverse weather conditions and reducing environmental impact through decreased fuel consumption. Currently, fishing voyage planning and route optimization rely heavily on the accumulated experience of skippers. Thus, it is a challenging task for inexperienced skippers. This paper presents an application of the reinforcement learning algorithm for fishing route planning and optimization. Reinforcement learning is a type of machine learning technique to learn an optimal behavior through a trial-and-error search in an environment to discover the sequence of actions under uncertainty that maximizes the defined reward. In this study, several optimization agents for different objectives, including maximizing catch, minimizing distance, maximizing catch while minimizing risk, etc., are trained from a programmed fishing environment to test out the capabilities of reinforcement earning for fishing route optimization. Future research can be developed by building upon the programmed fishing environment by incorporating a more realistic geographical map, uncertain fish stock distribution, a more sophisticated ship model, sound route risk function, fuel consumption function, etc. . Customized agents can be trained and further utilized as a decision support tool by skippers for their fishing trip planning and route optimization.
Forlag
ASMESitering
Zhu, Naseri, Dhar, Ashrafi: Reinforcement Learning for Fishing Route Planning and Optimization. In: 2024. Proceeding of ASME 2024 43rd International Conference on Ocean, Offshore and Arctic Engineering. Volume 6:Polar and Arctic Sciences and Technology; CFD, FSI, and AI, 2024. The American Society of Mechanical Engineers (ASME)Metadata
Vis full innførselSamlinger
Copyright 2024 The Author(s)