Next-Point Prediction for Direct Touch Using Finite-Time Derivative Estimation
summary
A three-step predictor is built that (1) estimates derivatives with a finite-time derivative estimator, (2) progressively disables prediction with an exponential waiting function when speed is low, and (3) applies two-stage smoothing. It runs on 6,000 benchmark strokes and a 9-participant user study, predicts up to 2–3× farther than prior methods, and supports realistic compensation up to about 48 milliseconds while noticeably reducing jitter, jumps and spring effects compared to industry and literature baselines.
AI-written summary · about
chapters (8)
- 0:00Intro
- 0:35Input Latency
- 1:53Touch Prediction
- 3:06Previous Contributions
- 4:34The Turbo Touch Predictor
- 7:50Benchmark Evaluation
- 10:49Subjective Feedback Experiment
- 14:02Summary
related talks
20mNext-Point Prediction Metrics for Perceived Spatial Errors
19mDoes Practice Make Perfect? Learning to Deal with Latency in Direct-Touch Interaction
19mInvestigating Effects of Post-Selection Feedback for Acquiring Ultra-Small Targets on Touchscreen
19mUsing High Frequency Accelerometer and Mouse to Compensate for End-to-end Latency in Indirect ...
19mPredicting Finger-Touch Accuracy Based on the Dual Gaussian Distribution Model
17mPointing Forward (updated)
23mThe Flat Finger: Exploring Area Touches on Smartwatches
18mPointing Forward