Development and Application of the Coverage Path Planning Based on a Biomimetic Robotic Fish

Abstract

This paper addresses the coverage path planning (CPP) and path-following challenges for an underwater biomimetic robotic fish, aimed at performing water quality monitoring in deep-sea net cages. First, a novel CPP strategy is proposed to minimize total path length, repetition rate, and turning frequency, thereby enhancing path coverage efficiency and reducing energy consumption. This strategy integrates the Deep Q-Network (DQN) method, a tailored reward function, and an RRT*-inspired rewrite mechanism. Second, a high-performance “LOS-PID” controller is developed to enable precise path following by the robotic fish. Finally, simulation experiments and field tests with the untethered robotic fish validate the effectiveness of the proposed CPP and path-following strategies, highlighting their practical applicability in aquaculture management. The proposed algorithm provides valuable insights into optimizing coverage path planning for underwater robotic systems in real-world scenarios.