Variance and Distribution Models for Steering Tasks
summary
Proposes a quadratic variance model that predicts movement-time variance as a quadratic function of index of difficulty with effectively zero linear term and explains between 78–97% of observed MT-variance across two datasets (10 and 12 participants). Combines steering law for mean and the variance model for scale, fits six candidate distributions with MLE/Bayesian methods, finds positively skewed fits outperform Gaussian and Gamma distribution performs best, validated by cross-validation, AIC/WAIC, and posterior predictive checks.
AI-written summary · about
related talks
18mModeling the Uncertainty in 2D Moving Target Selection
19mPredicting Finger-Touch Accuracy Based on the Dual Gaussian Distribution Model
20mNext-Point Prediction Metrics for Perceived Spatial Errors
26mReproducing Weekly Crash Patterns Using a Minimalist Driver Error Model
17mNext-Point Prediction for Direct Touch Using Finite-Time Derivative Estimation
46mQueueing Theory in Practice: Performance Modeling for the Working Engineer
48mRevisiting Performance Budgets
06mIntuitive Interaction with Motion Controls in a Tennis Video Game