Prof. Huashan Liu, Donghua University, China
Biography: Huashan Liu is a Professor at Donghua University for robotics, artificial intelligence, and mechatronics, a Senior Member of IEEE and a Member of the Intelligent Control and Systems Committee of the Chinese Command and Control Society. He received the B.E. degree in Mechanical Engineering from Wuhan University, Wuhan, China, and the Ph.D. degree in Mechatronics from Zhejiang University, Hangzhou, China in 2005 and 2010, respectively. Since 2010, he has been with the College of Information Science and Technology (CIST), Donghua University, Shanghai, China, where he founded and directed the Coexisting-Cooperative-Cognitive Robot Lab. During 2015-2016 he was a Visiting Professor with the Institute of Automatic Control Engineering (LSR), Technical University of Munich (TUM), Munich, Germany. Prof. Liu has published more than 70 articles in international academic journals such as Transaction series of IEEE and ASME, and has served as editorial board member or guest editor of more than 10 international academic journals in the field of robotics and artificial intelligence, invited speaker, session chair or program committee member of more than 30 international academic conferences, and reviewer of more than 50 international academic journals.
Title: From Dedicated to Universal: Robotic Motion Planning with AI
Abstract: Motion planning is a crucial foundation for robotic tasks. Conventional motion planning methods are highly dependent on inverse kinematics, which are extremely onerous for robots with redundant degrees of freedom. Moreover, the robots can merely execute fixed, pre-programmed, and hard-coded command sequences. Consequently, they cannot be generalized to different task scenarios. As an essential branch of artificial intelligence (AI), deep reinforcement learning (DRL) has shown great potential in realizing versatile robotic assignments that are difficult to be implemented by conventional methods. DRL equips robots the ability to optimize their behaviors by constantly interacting with the environment, which is a huge step in materializing autonomous and intelligent robotic tasks. This talk discusses the potential of AI in terms of DRL in realizing universal robotic motion planning, and also contributes to the intersection of DRL and robotics by summarizing the previous study on universal robotic motion planning by leveraging DRL, so as to provide an overview of this field and push the boundary of interdisciplinary research.