12 Steinicke Jerald 2010 RedirectedWalkingThresholds
The paper investigates how much users can be unknowingly redirected in immersive virtual environments (IVEs) without perceiving inconsistencies between their physical and virtual movements. It introduces a taxonomy of redirected walking techniques based on translation, rotation, and curvature gains applied to the virtual camera relative to tracked real‑world motion. Using three psychophysical two‑alternative‑forced‑choice (2AFC) experiments with 12 participants, the authors measured points of subjective equality (PSE) and detection thresholds (DT) for each gain type. Experiment 1 (rotation gains) showed a PSE of 0.96 and asymmetric DTs (0.67–1.24), indicating users can be turned up to 49 % more or 20 % less than perceived. Experiment 2 (translation gains) yielded a PSE of 1.07 and DTs (0.86–1.26), meaning distances can be down‑scaled by 14 % and up‑scaled by 26 % without detection. Experiment 3 (curvature gains) found a PSE near zero and DTs at ±0.045, corresponding to a minimum undetectable curvature radius of ~22 m. The authors discuss the implications of these thresholds for designing redirected‑walking locomotion interfaces, note asymmetries in perception, and outline future work on additional redirection methods, adaptation, and contextual factors. The paper also provides detailed hardware setup, participant demographics, and post‑experiment questionnaires on task difficulty, fear of collision, and simulator sickness.
Topics
Taxonomy of Redirected Walking Gains: Translation, Rotation, and Curvature
Defines the three core gain types (translation gain g_T[w], rotation gain g_R[u], curvature gain g_C[w]) within the Human Locomotion Triple framework and how they manipulate the virtual camera.
Psychophysical Methodology: 2AFC, PSE, and Detection Thresholds
Describes the experimental design using two‑alternative forced‑choice tasks, the calculation of point of subjective equality (PSE) and detection thresholds (DT) at 75%/25% response levels.
Experiment 1 – Rotation Gains: Asymmetric Perceptual Limits
Reports that subjects have a PSE of 0.96 for yaw rotation gains and can tolerate gains from 0.67 to 1.24 (≈‑20 % to +49 % deviation) without reliable discrimination.
Experiment 2 – Translation Gains: Underestimation of Virtual Distance
Finds a PSE of 1.07 for forward translation gains, with detectable limits at 0.86 and 1.26, indicating users can be redirected by –14 % to +26 % in distance.
Experiment 3 – Curvature Gains: Minimum Undetectable Arc Radius
Shows a near‑zero PSE for curvature gains and detection thresholds at ±0.045, corresponding to a minimum undetectable curvature radius of about 22 m.
Design Implications and Future Directions for Redirected Walking
Discusses how the measured thresholds inform VR locomotion interface design, notes asymmetries, potential for larger gains under task distraction, and outlines planned extensions (e.g., backward movement, adaptation, contextual cue effects).