REDUCING LATENCY IN A VIRTUAL REALITY-BASED TRAINING APPLICATION

Overall latency is the elapsed time from input of human motion to the immediate
response of the input in the display. Apparently, latency is one of the most frequently
cited shortcomings of current Virtual Reality (VR) applications. To compensate
latency, previous prediction mechanisms insert a complex mathematical algorithm,
which may not be appropriate for complex virtual training applications. More
complex VR simulations most likely will impose greater computation burdens and
resulted in the increase of latencies.
In order to overcome latency problem, this research is an attempt to suggest a new
prediction algorithm based on heuristic that could be used to develop a more effective
and general system for virtual training applications. The heuristic-based predictor
provides a platform to utilize the heuristic power of human along with the algorithmic
power, geometry accuracy of motion-planning programs and biomechanical laws of
human. Heuristic algorithm is an important module widely used for humanoid robots
and avatars in VR systems. However, to the best of the researcher's knowledge, the
heuristic approach has not been used as a single prediction algorithm for
compensating latency in virtual training systems.
In order to find out whether the new prediction algorithm is acceptable and possibly
could reduce latency, a fast synchronization squash-game simulation was selected as a
study source. This research analyzed the latencies of all subcomponents of this system
and designed prediction algorithm that allows high-speed interaction.
In measuring the performance on various prediction methods, this research also makes
a comparison in real tasks among 1) the heuristic-based prediction, 2) the Grey system
prediction and 3) the one without prediction using different sample rates. Findings
indicated that heuristic-based algorithm is an accurate prediction method to
compensate latency in virtual training. Apparently, heuristic-based prediction and
Grey system prediction are significantly better than the one without prediction. When
heuristic-based prediction and Grey system prediction were compared, heuristic-based
prediction was in fact a better predictor. Overall findings indicated that heuristicbased
prediction is efficient, robust and easier to implement.