Understanding Predictive Tracking and Motion Estimation Algorithms in Virtual Environments

Predictive tracking algorithms are in limited use in today's virtual reality systems. Choosing the appropriate prediction algorithms is a challenging task since there are many issues to consider when applying them in practical situations. Certain algorithms may perform better or worse depending on the underlying tracking system's sampling rate, noise variance and tracking technology (e.g., magnetic, acoustic, inertial, hybrid). The types of user motion, including head and hand, also play a significant role in determining what prediction algorithms to use. Another critical factor in prediction algorithm determination is the prediction time (i.e., how far one has to predict into the future). Some prediction algorithms may be more or less robust as prediction time increases. Almost all prediction algorithms contain one or more parameters that require tuning to optimize performance. Therefore, a significant aspect in determining what prediction algorithms to use is in adjusting an algorithm's parameter values. The adjustments are nontrivial in the sense that an optimal parameter setting for one type of user motion may not be optimal for another.

To aid VR developers and researchers in understanding and using predictive tracking and motion estimation algorithms in virtual and augmented reality, I have developed a testbed which contains user motion datasets, a variety of predictors, and a testing and analysis application. The first release of the testbed can be downloaded here.

Download the User Motion Data Repository
Download the Predictor Library and Testing Application

Publications

Julier, S., and LaViola, J. "On Kalman Filtering with Nonlinear Equality Constraints", IEEE Transactions on Signal Processing, 55(6):2774-2784, June 2007.

Julier, S. and LaViola J. "An Empirical Study into the Robustness of Split Covariance Addition (SCA) for Human Motion Tracking", In the Proceedings of the 2004 American Control Conference, IEEE Press, 2190-2195, June 2004. (Note the published version of this paper has some notational errors. This version is correct.)

LaViola, J. "A Comparison of Unscented and Extended Kalman Filtering for Estimating Quaternion Motion", In the Proceedings of the 2003 American Control Conference, IEEE Press, 2435-2440, June 2003.

LaViola, J. "A Testbed for Studying and Choosing Predictive Tracking Algorithms in Virtual Environments", In the Proceedings of Immersive Projection Technology and Virtual Environments 2003, ACM Press, 189-198, May 2003.

LaViola, J. "Double Exponential Smoothing: An Alternative to Kalman Filter-Based Predictive Tracking", In the Proceedings of Immersive Projection Technology and Virtual Environments 2003, ACM Press, 199-206, May 2003.

LaViola, J. "An Experiment Comparing Double Exponential Smoothing and Kalman Filter-Based Predictive Tracking Algorithms", In the Proceedings of Virtual Reality 2003, IEEE Press, 283-284, March 2003.
Joseph LaViola
Last modified: Fri Jul 2 00:04:19 EDT 2004