Kalman Filter for Tracking Hand Gestures
We propose to develop new Kalman filter-based algorithms for a hand gesture detection and tracking. We currently study detection of static gestures using motion-capture cameras with unlabeled markers. The new Kalman filter will enable tracking of multiple markers simultaneously even when some markers are occluded and will allow for marker identification in real-time. This research will enable tracking a very fine motion of individual thumb and fingers simultaneously, which is fundamentally different than presently available gesture detection systems such as Microsoft Kinect. The system will be able to recognize both static and dynamic hand gestures in three dimensional space. We will consider a single and dual Kalman filters that track thumb and fingers and also enforce rigidity at the back of a hand. Such novel design will improve stability and accuracy of hand, thumb and fingers position estimations. We propose to conduct simulations and live experiments in order to verify novel algorithms and overall technology development. The Kalman filter algorithms for hand gesture tracking will be implemented at our motion caption laboratory. This technology also has broad applications in multiple target tracking, smart vehicle control, self-controlled cars, and many others. This proposed project is expected to lead to more about robust and user-friendly control methods that will impact NASA in-vehicle operations and control, improve safety of their personnel, and advance NASA missions in space.
Principal Investigator: Selmic, Rastko -- Electrical Engineering
|Start Period: 00/00/0000
||End Period: 00/00/0000
No Affiliated People