Modeling Pointing for 3D Target Selection in VR
summary
Motion capture from seven hand, arm, and body locations across 1755 trials at 27 targets spaced one meter apart yields models that predict distant pointing with 87% classification accuracy and a mean error of 23.56 cm. Three feature sets are compared: a final-pose set (positional, categorical, shoulder flexion) gives the best accuracy, a full-movement set uses third-order polynomial coefficients plus velocity/acceleration and performs worse but above chance, and a head‑wrist‑hand subset matches full-body results.
AI-written summary · about
related talks
20mModeling the Impact of Depth on Pointing Performance
22mPinpointing: Precise Head- and Eye-Based Target Selection for Augmented Reality
06mOverthere : A Simple and Intuitive Object Registration Method for an Absolute Mid-air Pointing ...
11mExtendlibur: Dynamic Haptic Retargeting for Length-Mismatched Proxies in Co-Located VR
06mCoolMoves: User Motion Accentuation in Virtual Reality
11mStabilizAR: Enabling Hands-Free Head Pointing while Mobile
19mVirtual Replicas for Remote Assistance in Virtual and Augmented Reality
20mSelection-based Text Entry in Virtual Reality