F-RDW: Redirected Walking With Forecasting Future Position.

Area of Science:

• Virtual Reality
• Human-Computer Interaction
• Computer Science

Background:

• Existing Redirected Walking (RDW) methods improve VR experiences by predicting user movement to minimize resets.
• Current RDW techniques often require specific virtual environment layouts or user walking patterns, limiting their applicability.

Purpose of the Study:

• To propose F-RDW, a flexible mechanism that forecasts user movement without assumptions and integrates this information with existing RDW methods.
• To enhance the universal application and performance of RDW in diverse VR scenarios.

Main Methods:

• Developed an LSTM-based model to predict future user positions using spatial and eye-tracking data.
• Integrated predicted future information into established RDW algorithms (MPCRed, S2C, TAPF, ARC).

Main Results:

• Simulations and user studies demonstrated the effectiveness of F-RDW in reducing reset occurrences.
• The mechanism significantly increased the distance traveled between resets, improving overall RDW performance.
• F-RDW proved beneficial in small physical spaces and complex virtual environments.

Conclusions:

• F-RDW effectively leverages predicted user movement to overcome limitations of existing RDW methods.
• The proposed mechanism offers a more universally applicable and performant solution for enhancing VR experiences through Redirected Walking.