A STATISTICAL MODEL OF LOOMING DETECTION

Looming detection plays a substantial role in headway control. One philosophy of designing car following systems is to base their characteristics on those of a highly attentive experienced driver. This requires a model of a driver's perceptual limitations in judging distance changes to a preceding vehicle. Experimental data is collected to explore these limitations and initiate development of a statistical model of looming detection. Three processes are distinguished in the model: (i) target size is perceived with a multiplicative uncertainty on the visual angle subtended by the target; (ii) during initial exposure of a stationary target, this uncertainty decreases as the result of a noise integration process, while simultaneously; (iii) the uncertainty increases due to a memory fading process. For a wide variety of looming conditions, observed and model-predicted detection times (up to six seconds) as well as their distributions are shown to match well. The model computes the time it takes for the target's changing visual angle to differ significantly from the subject's noisy dynamic internal representation of the initial target size. Furthermore, the effect of radial flow is shown to increase perceptual noise, strengthen the benefit of prolonged noise integration, and increase memory fading. (A) For the covering abstract see ITRD E106152.

  • Availability:
  • Corporate Authors:

    Elsevier

    The Boulevard, Langford Lane
    Kidlington, Oxford  United Kingdom  OX5 1GB
  • Authors:
    • Boer, E R
  • Publication Date: 1999

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 00799764
  • Record Type: Publication
  • Source Agency: Transport Research Laboratory
  • ISBN: 0-08-043671-4
  • Files: ITRD
  • Created Date: Oct 6 2000 12:00AM