Interactive Media Systems, TU Wien

VRGroup - Gait Estimation from sparse Motion Data

Master Thesis, Bachelor Thesis, or Practicum

About This Topic

In a Virtual Reality scenario the animation of the user's avatar contributes a lot to the feeling of immersion. Employing a full body motion capture solution can make the system unnecessarily complex and expensive, especially if precise motion animation is not crucial. Usually there is already a lot of motion data available from other sources, for example head position and rotation, movement speed, hip orientation,.... This could allow an estimation of the user's most probable pose and walking phase.

Additional Information

There is already initial work done to start from, estimating the arm pose from Leap data with FABRIK in Unreal.

Suggested Setup:

-) Unreal Engine

-) HTC Vive

-) Perception Neuron Motion Suit

-) Leap Motion

-) Cyberith Virtualizer