Marker-based based motion capture is the prevalent technique for estimating human motion. A common problem with the approach is the occlusion and mis-labeling of the markers; typically the data requires tedious manual cleaning in post processing. We present a constrained extended Kalman filter method that estimates full body human motion in real time and handles missing and mis-labeled markers. The approach is validated on two datasets and is shown to produce comparable results to using manually cleaned data. The constrained estimator ensures realistic human joint trajectories that satisfy kinematic limits.
CITATION STYLE
Joukov, V., Lin, J. F. S., Westermann, K., & Kulić, D. (2020). Real-Time Unlabeled Marker Pose Estimation via Constrained Extended Kalman Filter. In Springer Proceedings in Advanced Robotics (Vol. 11, pp. 762–771). Springer Science and Business Media B.V. https://doi.org/10.1007/978-3-030-33950-0_65
Mendeley helps you to discover research relevant for your work.