基于神经网络的机器人学习与控制：回顾与展望

机器人因其高效的感知、决策和执行能力，在人工智能、信息技术和智能制造等领域中具有巨大的应用价值。目前，机器人学习与控制已成为机器人研究领域的重要前沿技术之一。各种基于神经网络的智能算法被设计，从而为机器人系统提供同步学习与控制的规划框架。首先从神经动力学(ND)算法、前馈神经网络(FNNs)、递归神经网络(RNNs)和强化学习(RL)四个方面介绍了基于神经网络的机器人学习与控制的研究现状，回顾了近30年来面向机器人学习与控制的智能算法和相关应用技术。最后展望了该领域存在的问题和发展趋势，以期促进机器人学习与控制理论的推广及应用场景的拓展。

Learning and Control of Robots Based on Neural Networks: Review and Outlook

Robots have great application value in the fields of artificial intelligence, information technology, and intelligent manufacturing due to their efficient perception, decision-making, and execution capabilities. At present, robot learning and control have become one of the most critical frontier technologies in the field of robotics. Meanwhile, different intelligent algorithms based on neural networks have been designed to provide a planning framework for synchronous learning and control of robot systems. Specifically, the research status of neural-network-based robot learning and control is reviewed from four aspects:neural dynamics (ND) algorithms, feedforward neural networks (FNNs), recurrent neural networks (RNNs), and reinforcement learning (RL). The intelligent algorithms and related application technologies utilized for robot learning and control in the past three decades are reviewed in detail. Finally, the remaining challenges and development trends in this field are provided to promote the development of robot learning and control theory and the extension of application scenarios.