【作者】 赵茜妍； 许佳； 李莉；

【Author】 Zhao Xiyan;Xu Jia;Li Li;College of Electronics and Information Engineering, Tongji University;

【机构】 同济大学电子与信息工程学院；

【摘要】 协同避障任务是指在复杂动态环境下,多个目标智能体自主决策避开障碍物和其他智能体,安全高效地到达目标位置。本文概括了深度强化学习的基本原理,分析了基于值函数和基于策略梯度两类算法的特点。在此基础上梳理了基于深度强化学习的多智能体协同避障方法的主要工作和研究进展,并探讨了其面临的挑战和未来发展方向。更多还原

【Abstract】 Collaborative obstacle avoidance task refers to that in complex dynamic environment,multiple target agents make independent decisions to avoid obstacles and other agents, and reach the target position safely and efficiently. This paper briefly summarizes the basic principles of deep reinforcement learning, and analyzes the characteristics of two kinds of algorithms based on value function and strategy gradient. On this basis, the main work and research progress of multi-agent cooperative obstacle avoidance method based on deep reinforcement learning are summarized, and its challenges and future development are discussed.更多还原