इसका टेक्स्ट मैसेज भेजे: A Reinforcement Learning Approach Based on Automatic Policy Amendment for Multi-AUV Task Allocation in Ocean Current