This study aims to solve the issue of the safe navigation of autonomous underwater vehicles (AUVs) in an unknown underwater environment. AUV will encounter canyons, rocks, reefs, fish, and underwater vehicles that threaten its safety during underwater navigation. A game-based soft actor–critic (GSAC) path planning method is proposed in this study to improve the adaptive capability of autonomous planning and the reliability of obstacle avoidance in the unknown underwater environment. Considering the influence of the simulation environment, the obstacles in the simulation environment are regarded as agents and play a zero-sum game with the AUV. The zero-sum game problem is solved by improving the strategy of AUV and obstacles, so that the simulation environment evolves intelligently with the AUV path planning strategy. The proposed method increases the complexity and diversity of the simulation environment, enables AUV to train in a variable environment specific to its strategy, and improves the adaptability and convergence speed of AUV in unknown underwater environments. Finally, the Python language is applied to write an unknown underwater simulation environment for the AUV simulation testing. GSAC can guide the AUV to the target point in the unknown underwater environment while avoiding large and small static obstacles, canyons, and small dynamic obstacles. Compared with the soft actor–critic(SAC) and the deep Q-network (DQN) algorithm, GSAC has better adaptability and convergence speed in the unknown underwater environment. The experiments verifies that GSAC has faster convergence, better stability, and robustness in unknown underwater environments.

No abstract

Reinforcement learning (RL) is known for its efficiency and practicality in single-agent planning, but it faces numerous challenges when applied to multi-agent scenarios. In this paper, a Super Sampling Info-GAN (SSIG) algorithm based on Generative Adversarial Networks (GANs) is proposed to address the problem of state instability in Multi-Agent Reinforcement Learning (MARL). The SSIG model allows a pair of GAN networks to analyze the previous state of dynamic system and predict the future state of consecutive state pairs. A multi-agent system (MAS) can deduce the complete state of all collaborating agents through SSIG. The proposed model has the potential to be employed in multi-autonomous underwater vehicle (multi-AUV) planning scenarios by combining it with the Soft Actor–Critic (SAC) algorithm. Hence, this paper presents State Super Sampling Soft Actor–Critic (S4AC), which is a new algorithm that combines the advantages of SSIG and SAC and can be applied to Multi-AUV hunting tasks. The simulation results demonstrate that the proposed algorithm has strong learning ability and adaptability and has a considerable success rate in hunting the evading target in multiple testing scenarios.