Abstract

The paper summarizes a project to increase the sense of presence within a driving simulator while interacting with autonomous traffic. The project sought to model natural variations in ambient traffic to emulate identifiable driving styles for different categories of driver. Probability distributions combined with decision histories were employed to characterize speed choice while providing a mechanism for introducing temporal and spatial variation in speed changes. These efforts produced “virtual personalities” representing different categories of ambient traffic including generic, male, female, old, drunk, aggressive, cautious, and fatigued. A user evaluation of the ambient traffic concluded that naturalistic variation in behavior can significantly contribute to the subjective realism of the interaction with traffic simulation.

This content is only available as a PDF.