Evaluating the Performance of Hand-Based Probabilistic Text Input Methods on a Mid-Air Virtual Qwerty Keyboard

Area of Science:

• Human-Computer Interaction
• Virtual Reality
• Usability Studies

Background:

• Virtual reality (VR) headsets with integrated hand-tracking offer potential for mid-air text entry.
• Virtual keyboards can mimic physical Qwerty layouts, allowing users to utilize existing typing skills.
• Challenges include lack of haptic feedback, unconstrained motion, and tracking inaccuracies, which degrade input accuracy.

Purpose of the Study:

• To explore error-tolerant probabilistic hand-based input methods for mid-air virtual Qwerty keyboards.
• To evaluate the performance of hand-based text input using both gesture and touch typing paradigms in VR.
• To identify methods for improving text entry accuracy and efficiency in VR environments.

Main Methods:

• Development and implementation of probabilistic, error-tolerant algorithms for hand-based input.
• Conducting three user studies to assess text entry performance.
• Investigating both gesture and touch typing interaction styles on a virtual Qwerty keyboard.

Main Results:

• Demonstrated typical text entry rates between 20 and 30 words per minute (wpm).
• Achieved average peak entry rates ranging from 40 to 45 wpm.
• Validated the effectiveness of error-tolerant methods in mitigating accuracy degradation.

Conclusions:

• Error-tolerant probabilistic methods significantly enhance text entry on mid-air virtual keyboards.
• Hand-based input in VR is a viable method for text entry, with performance comparable to some physical interfaces.
• Further research can optimize these methods for even greater efficiency and user experience in virtual environments.