Prof. Shihui Guo

Xiamen University, China

​

Title: Physics-based Motion Controllers for Virtual Characters with Deep Neural Network

Designing the controller to actuate the virtual character is a challenging task given the high dimensions of both character states and controls. Flexible user control is normally required in interactive applications (virtual and augmented reality), which further adds to the difficulty of constructing a motion controller for the virtual characters. Although deep neural network showed its potential in tackling this problem, existing works based on neural networks normally require manual design of the network architecture, which requires professional experience and trial-and-error process. This talk will first give an overview of existing works in using deep neural network and reinforcement learning for the task of simulating locomotion skills of virtual characters, and then present my most recent findings in finding the optimal network structure

​

​

BIO

Shihui Guo is an associate professor with School of Informatics, Xiamen University. He obtained his bachelor from Yuanpei College, Peking University (2010), China and Ph.D. from National Centre for Computer Animation, Bournemouth University, UK (2015). His Ph.D. supervisors are Prof. Jianjun Zhang and Assoc Prof. Jian Chang. During his Ph.D, he also visited Prof. Wencheng Wang from State Key Laboratory of Computer Science in Institute of Software, Chinese Academy of Science. After that he joined Institute of Media Innovation in Nanyang Technological University, under the supervision of Prof. Nadia Thalmann, before moving back to China. His research interests including artificial intelligence, computer vision and graphics. He published ~30 research papers in international journals and conferences, including IEEE transactions on cybernetics, Computer Graphics Forum etc.