报告内容摘要：
Fish are excellent swimmers with high swimming efficiency and perfect manoeuvring abilities. Designing a robotic fish with fish-like swimming behaviour in a dynamic environment is such a challenging task, that it requires a full understanding of hydrodynamics, fish propulsion mechanisms, and the approaches for mimicking the real fish swimming. Our team of University of Essex has developed the first autonomous robotic fish. This research talk aims to give an overview of my research at University of Essex where I developed: (i) a control architecture for autonomous control, (ii) basic fish-like swimming ability, (iii) behaviour-based intelligent control, and (iv) online learning algorithms.
Due to the complexity of multi-joint motion control and dynamic changes in the real world, a hybrid control architecture is designed with layered control features.  It consists of three layers: (i) a swim pattern layer, which is designed for the low-level motion control of the multiple joints of our robotic fish; (ii) a behaviour layer, which processes sensory information for reactive control in order to handle uncertainty in the environment; and (iii) a cognitive layer, which plans tasks and coordinates multiple behaviours. A novel modelling methodology is proposed for the design of fish swim patterns, including three steady and one unsteady swim patterns. This provides our robotic fish with the capability to produce 3D fish-like swimming, which is unique. The behaviour layer is implemented using a fuzzy logic controller, which can effectively deal with noise and uncertainty. Four individual behaviours are designed to make the fish avoid obstacles, maintain a desired depth, follow a wall, and wander around. These behaviours are merged together using a coordination method that is configured by the cognitive layer. Sample-based Policy Gradient Reinforcement Learning (PGRL) is adopted to optimise control parameters in the swim pattern layer and behaviour layer. Q-learning is used to find a mapping between discrete states and actions in the cognitive layer. Experimental results from simulation and real robotic fish prove the feasibility and robustness of the proposed approach.
 
报告人简介：
刘晋东，男，博士，英国伦敦帝国理工大学研究员，主要研究领域在仿生移动机器人，包括自然人机交互，仿生机器鱼，柔性机器人，医疗及手术机器人。他于2003年毕业于中国科学院自动化研究所，获得控制理论与控制工程硕士学位，同年获得全额奖学金，赴英国埃塞克斯大学攻读计算机博士学位，2007年毕业。博士期间成功研制了世界第一条全自主水下仿生机器鱼。2008年至2010年期间于英国桑德兰大学担任副研究员，主要研究基于人脑听觉神经系统的机器人声音识别与定位系统。2010年加入英国伦敦帝国理澳门太阳娱乐网站官网哈姆林医学机器人研究中心，主要从事自然人机语音交互，传感器网络，医疗保健机器人的研究。在SCI及EI检索的期刊与会议发表论文40余篇，被引用500余次，文章主要发表于Neurocomputing， Bionics Engineering， Journal of Neural Network World， International Journal of Automation and Computing，IROS 及ICRA机器人期刊及会议。在2013年，他于Bionics Engineering上关于机器鱼的论文被评为引用次数最多的论文。他目前担任多个国际会议的委员会成员，以及多个IEEE期刊审稿人。他于2008，2009， 2013 及2014年被收录于Marquis世界名人录。
 
Dr. Liu is a research fellow at the Hamlyn Centre for Robotic Surgery, Imperial College London. He is interested in fields related to Biologically Inspired Mobile Robotics, mainly include natural human-robot interaction, biomimic robotic fish and compliant manipulator for healthcare and surgery robotics.  He has a PhD from the University of Essex where he focused on biologically inspired autonomous robotic fish. He successfully built the first autonomous robotic fish. Between 2008 to 2010, he shifted his interests to the human auditory system. He developed a computational mammalian auditory system applied to the sound perception on mobile robotics at University of Sunderland with collaboration of University of Newcastle. In 2010, he moved to London and joined the Hamlyn Centre, Imperial College London. Now he is focusing on natural human-robot speech interaction, pervasive sensing and healthcare mobile robots. His articles have been published in Neurocomputing, Journal of Bionic Engineering, Journal of Neural Network World and International Journal of Automation and Computing. He is a member of IEEE and reviewer of conferences and journals of IEEE and Springer including IEEE Trans. of Neural Network, IROS and ICRA, etc. In 2012, he won the “Best Poster Award” in 9th Int. Conf. of Body Sensor Network.  In 2013, his paper had been evaluated as “Most cited Article in 2012” in J. of Bionic Eng. He is listed in the Marquis Who’s Who in the world in 2008, 2009, 2013 and 2014.
 
欢迎广大老师和同学们参加！