A Steering Model for On-Line Locomotion Synthesis Abstract For
applications such as video games and virtual walk-throughs, on-line
locomotion control is an important issue. In general, the user prescribes a
sequence of motions one by one while providing an input trajectory. Since the
input trajectory lacks in human characteristics, one may not synthesize
quality motions by blindly following it. In this paper, we present a novel
data-driven scheme for transforming a user-prescribed trajectory to a human
trajectory in an on-line manner. As preprocessing, we analyze example motion
data to extract human steering behavior. At run-time, the input trajectory is
refined to reflect the steering behavior. Together with an existing on-line
motion synthesis system, our scheme forms a feedback loop, in which the user
effectively specifies an intended human trajectory. Paper & Demo |
|