More about HKUST
Character Motion Representation based on Spatial Relationships
Speaker: Dr. Taku KOMURA School of Informatics University of Edinburgh Title: "Character Motion Representation based on Spatial Relationships" Date: Thursday, 22 July 2010 Time: 3:00pm - 4:00pm Venue: Room 3311 (via lifts 17/18), HKUST Abstract: Close interactions, not necessarily with any contacts, between different body parts of single or multiple characters or with the environment are common in computer animation and 3D computer games. Yoga, wrestling, dancing and moving through a constrained environment are some examples. Existing scene representations have a fundamental limitation in handling such close interactions. Currently, a motion is typically described in terms of joint angles and kinematic constraints such as contacts. With this representation, automatically computing a valid motion requires randomized exploration and significant computation for collision detection. The animator also needs to shoulder the burden of specifying all the kinematic constraints in advance. From the animator's perspective, this is impractical and not conductive to manual editing. Competitive automatic solutions require an effective representation that allows the extraction of spatial relationships from existing motion data and synthesis of new animations that preserve these relationships. Such a representation will not only allow quantitative evaluation of the way different body parts are interacting, but also facilitate qualitative characterization of scene semantics. Our research group has been exploring new representations which considers the spatial relationships for describing the interactions of multiple characters or characters in a constraint environment. In this talk, I introduce two of the proposed methods. The first representation is called the topology coordinates that describes body parts are twisted around each other. The second representation is called the interaction mesh, which describes which body parts are in close proximity with others. Using these representations, we can easily edit or retarget human motions while preserving the context of the scene. The methodologies have a wide range of applications in fields of computer animation, pattern recognition and robotics. We will first briefly cover the overview of each method, and then show some demos of applying them to character motion synthesis. ******************* Biography: Taku Komura is currently a Lecturer in School of Informatics at the University of Edinburgh. Before joining University of Edinburgh on 2006, he worked as an Assistant Professor in City University of Hong Kong (2002-2006), and as postdoctoral researcher at RIKEN, Japan. He receivied his PhD (2000), MSc (1997) and BSc (1995) in Information Science from the University of Tokyo. His research interests include human motion analysis and synthesis, physically-based animation and topology-based modelling. His research area covers computer graphics, robotics and biomechanics.