Purchase this article with an account.
Borja Aguado Ramirez, Joan López-Moliner; Active sampling of the optic flow to predict time-to-contact. Journal of Vision 2020;20(11):795. doi: https://doi.org/10.1167/jov.20.11.795.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
How we are able to estimate time-to-contact (TTC) in a 3D parabolic trajectory based on sparse depth reliable visual cues is still a challenging research issue. So far, the ecologic approach has been predominant relying on Tau and other optical variables to estimate TTC. However, Tau cannot account for the estimation of time-to-contact in different conditions such as: high lifted balls, accelerated objects or trajectories in a non collision course with the observer. We created a model that can accurately predict TTC in these situations, making use of contextual information such as: known size, gravitational acceleration and estimates of rate of the elevation angle mainly. In fact, the model predicts different outcomes depending on the portion of parabola that is visible. The aim of this study was to test the correspondence of our model’s predictions with observers’ estimates. To do so, we designed a task in which participants, wearing a HMD (HTC @90Hz/eye), had to time the moment a ball in a parabolic path (3 or 3.5 s flight time) returned at eye-level with a button press. We used five trajectories for which the model made accurate predictions at different times of the flight. The ball was visible during the first 300 ms. after launch. After the target was launched, participants had to look at an arrow on the floor that would indicate which commander they would use to perform the temporal judgment. While looking at the floor, they were free to choose when to look for the flying ball. As soon as they looked up, the ball was visible for another period of 400 ms. The results indicate that our model can accurately predict the observer’s temporal estimates mainly based on a measure of the vertical speed within both visibility windows with the second one depending on participants' action.
This PDF is available to Subscribers Only