Reinforcement Learning Based Underwater Wireless Optical Communication Alignment for Autonomous Underwater Vehicles

With the developments in underwater wireless optical communication (UWOC) technology, UWOC can be used conjunction with autonomous vehicles (AUVs) for high-speed data sharing among vehicle formation during exploration. A beam alignment problem arises due to transmission range, external disturbances and noise, uncertainties AUV dynamic model. In this article, we propose an acoustic navigation method guide process without requiring directors, light intensity sensors, and/or scanning algorithms as previous research. The AUVs need stably maintain a specific relative position orientation establishing link. We model partially observable Markov decision (POMDP) that takes manipulation, navigation, energy consumption of into account. However, finding efficient policy POMDP under high partial observability environmental variability is challenging. Therefore, successful optimization, utilize soft actor–critic reinforcement learning algorithm together AUV-specific belief updates reward shaping based curriculum learning. Our approach outperformed baseline approaches simulation environment successfully performed from one another on real Tri-TON 2.