A Literature Analysis-Based Study on Advances in Underwater Multi-Robot Pursuit-Evasion Problems

• Published in: 水下无人系统学报 Vol. 33; no. 3; pp. 484 - 494
• Main Authors: Zhenkun LEI, Mingzhi CHEN, Daqi ZHU
• Format: Journal Article
• Language: Chinese
• Published: Science Press (China) 01.06.2025
• Subjects: model predictive control; multi-robot systems; pursuit-evasion; reinforcement learning; underwater robot
• ISSN: 2096-3920
• DOI: 10.11993/j.issn.2096-3920.2025-0032

Investigating the applications and challenges of multi-robot pursuit-evasion problems in underwater environments holds significant importance for enhancing the autonomous decision-making and collaborative capabilities of underwater robot systems. By searching the Web of Science Core Collection database, over 2 200 relevant literatures published between 2004 and 2024 were screened, and a comprehensive analysis was conducted on the definition of pursuit-evasion problems, research status, intelligent pursuit-evasion methods, and their applications in underwater environments. The principles, advantages, disadvantages, and applicability of four intelligent pursuit-evasion methods, including reinforcement learning, model predictive control, Apollonius circle, and artificial potential field, were analyzed in depth. The study reveals that reinforcement learning optimizes strategies through training to adapt to complex environments but suffers from a long training cycle; model predictive control formulates strategies