Purchase this article with an account.
Daniel R. Berger, Jörg Schulte-Pelkum, Heinrich H. Buelthoff; How to simulate realistic forward accelerations on a 6dof motion platform. Journal of Vision 2004;4(8):796. doi: 10.1167/4.8.796.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
How are visual and physical motion cues integrated in self-motion perception? We performed a psychophysical study to test how forward accelerations can be realistically simulated. Participants were seated on a 6 dof Stewart motion platform and viewed a computer-generated visual scene on a projection screen (54 ×40.5 ). The visual scene consisted of a randomly structured ground plane and sky. Observers were told that eye height was always 2m above ground. In each of the 180 trials, participants experienced a brief simulated forward acceleration (4s ramp, followed by 2s of constant acceleration) that was presented both as platform motion and motion within the visual scene. After the acceleration, the screen went dark and the platform returned to zero in 6s. Participants used a joystick to indicate the realism of the forward acceleration experienced. They were explicitly told to give high ratings for trials in which they convincingly felt moving forward in accordance with the visual stimulus, and low ratings in trials in which they noticed conflicts. In each trial, stimuli were chosen randomly within fixed intervals for all six varied parameters: visual forward accelerations (0–1.5 m/s^2), platform backwards pitch (0–15 ), brief forward translations of the platform (0–0.5m in 4s), ratio of acceleration/deceleration durations for the translations (0.11–1.5), and up/down noise simulating ground roughness (0–7cm) using low-pass filtered noise (cosine window, 0.3–1s). Multiple hierarchical regression analyses performed for each subject revealed that only two parameters had a clear influence on the ratings: higher platform pitches and higher visual accelerations induced a better impression of being accelerated forward. Brief forward translations of the platform and bumps increased believability for some observers. Interestingly, during sensory conflict between canal and visual/otolith cues, the latter dominated, as predicted by a Bayesian model that holds vision as more reliable.
This PDF is available to Subscribers Only