Time to Contact Estimation in Virtual Reality

INTRODUCTION: Virtual Reality (VR) headsets that can display binocular stimuli are now widely available to the general public. Many tasks and scenarios presented in VR involve dynamic visual objects. Yet, no systematic studies have been conducted to investigate motion perception in VR. We conducted three experiments to quantify the perception of looming motion in VR, using judgment of time to contact (TTC), and to evaluate the effectiveness of interventions to improve accuracy of TTC estimates. METHODS: Observers viewed a virtual baseball stadium from the perspective of a batter standing in the batter's box, using an Oculus Rift VR headset. Simulated balls were pitched at the observer. In each trial the ball was visible for a brief duration and moved at one of four constant speeds, between 20 and 83 mph. The ball trajectory was either fully visible or disappeared after 1/4, 1/2, and 3/4 of the trajectory. Observers judged TTC by button press. RESULTS: Observers (n=10) generally underestimated ball speed in both VR and non-VR settings. TTC accuracy systematically improved with increasing presentation duration and decreasing speed in both settings. Next we investigated three interventions (in n=19) to improve the accuracy of TTC estimates in VR. These manipulated different perceptual cues that inform speed perception: (1) The speed of the model ball increased. (2) The size of the model ball increased as it approached the observer, providing monocular cues of increased speed. (3) The vergence angle increased, providing binocular cues of increased speed. All three interventions improved TTC estimation accuracy, with better correction at lower speeds. CONCLUSION: The findings indicate that there is systematic underestimation of speed in VR, which can be effectively corrected by different interventions. Meeting abstract presented at VSS 2017