An Efficient Multi-AUV Cooperative Navigation Method Based on Hierarchical Reinforcement Learning

Abstract

Positioning errors introduced by low-precision navigation devices can affect the overall accuracy of a positioning system. To address this issue, this paper proposes a master-slave multi-AUV collaborative navigation method based on hierarchical reinforcement learning. First, a collaborative navigation system is modeled as a discrete semi-Markov process with defined state and action sets and reward functions. Second, trajectory planning is performed using a hierarchical reinforcement learning-based approach combined with the polar Kalman filter to reduce the positioning error of slave AUVs, realizing collaborative navigation in multi-slave AUV scenarios. The proposed collaborative navigation method is analyzed and validated by simulation experiments in terms of the relative distance between the master and slave AUVs and the positioning error of a slave AUV. The research results show that the proposed method can not only successfully reduce the observation and positioning errors of slave AUVs in the collaborative navigation process but can also effectively maintain the relative measurement distance between the master and slave AUVs within an appropriate range.