VR latency study: accuracy degrades before you feel the lag
New research shows even tiny hand-to-virtual-hand delays ruin precision first.
A new study from the University of Toronto (Wang, Sabiston, Welsh) systematically examined interaction latency in virtual reality. Participants performed manual pointing tasks while delays of 0 to 500ms were imposed between their physical hand movement and the rendered virtual hand. The researchers measured endpoint error, movement time, endpoint variability, and throughput, comparing VR performance to a physical baseline.
The results showed a clear dissociation: endpoint accuracy degraded even at the shortest imposed delays, while movement time increased only at longer delays. This means spatial measures are far more sensitive to latency than temporal ones. The authors conclude that traditional HCI evaluations relying solely on movement time or throughput can miss early disruption. For VR applications requiring precise manual actions—such as training simulations, remote surgery, or design tools—developers must prioritize spatial accuracy metrics to catch latency issues before users notice sluggishness.
- Endpoint error increased even at the shortest imposed delays (e.g., 50ms) between physical and virtual hand movement.
- Movement time only degraded significantly at longer delays (around 300-500ms), masking early latency issues.
- HCI evaluations should include spatial accuracy measures (like endpoint error) alongside temporal measures to fully assess VR latency impact.
Why It Matters
VR designers must measure spatial accuracy, not just speed, to detect latency problems early in interactive tasks.