September 2018
Volume 18, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2018
Testing models of speed control in 1D pedestrian following
Author Affiliations
  • Jiuyang Bai
    Department of Cognitive, Linguistic and Psychological Science, Brown University
  • William Warren
    Department of Cognitive, Linguistic and Psychological Science, Brown University
Journal of Vision September 2018, Vol.18, 1034. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Jiuyang Bai, William Warren; Testing models of speed control in 1D pedestrian following. Journal of Vision 2018;18(10):1034.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Global patterns of crowd behavior are believed to result from local interactions between pedestrians. Many studies have investigated the local rules of interaction, but how a pedestrian controls walking speed when following a leader remains in dispute. The present study experimentally tested six speed control models from the pedestrian and car following literature. These dynamical models control the follower's acceleration based on the leader's distance (distance model), speed difference (speed model), a combination of speed and distance (speed-based distance model, ratio model, linear model), or visual angle (optical expansion model). Previously, Rio, Rhea, & Warren (2014) reported evidence consistent with several models; here we dissociate them by testing a wider range of initial distance conditions. A participant (N=10) walked in a virtual environment while wearing an Oculus CV1 HMD, and head position was recorded (sampled at 60 Hz). They were asked to follow a virtual moving target pole for 12m, which changed speed after 2-3s. The target's initial distance (1, 4, 8m), initial speed (0.8, 1.2 m/s), and change in speed (-0.3, 0, +0.3 m/s) were randomly varied on each trial. All variables had significant effects on the participant's final speed and distance (p< .0001). Each model was fit to the participants' time series of speed using a Monte Carlo cross-validation procedure with 100 repetitions and a 75%/25% training/test split, then parameters were fixed. The linear model (four free parameters) had the highest correlation with follower's speed, but the optical expansion model (one free parameter) had the lowest RMS error in speed in all conditions; the other models exhibited an increase in RMS error at longer distances. The results imply that pedestrians directly control their walking speed by accelerating to cancel the optical expansion/contraction of the leader, rather than relying on the leader's distal speed or distance.

Meeting abstract presented at VSS 2018


This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.