Pedestrian Safety: Virtual Reality Simulator Development and Validation for Analysis of Alternative Safety Technologies

Angulo, Austin, Civil Engineering - School of Engineering and Applied Science, University of Virginia
Smith, Brian, EN-Eng Sys and Environment, University of Virginia

Rising rates of pedestrian fatalities is an urgent concern in the field of transportation. Both the National Highway Traffic Safety Administration (NHTSA) and Virginia Department of Transportation (VDOT) report gradual decreasing in pedestrian related crashes; however, they both report increasing pedestrian fatality rates. NHTSA reported a 35% increase in pedestrian fatalities nationwide between 2008 and 2017 and VDOT reported a 19% increase in pedestrian fatality rates between 2012 and 2018 in the state of Virginia.

Efforts to understand pedestrian behavior and safety have traditionally relied on real world observation methods; however, these methods are time consuming, costly, and unrealistic. With respect to motorists, driving simulators have become more sophisticated over the years and are now used as tools for understanding driver behavior and safety in realistic conditions. Efforts in creating virtual environments have been developed and tested for use in understanding non-motorized traveler behavior and safety, though, previous technologies have struggled to provide realistic and immersive environments due to the greater degree of freedom pedestrians wield over motorists.

The recent advancement of virtual reality (VR) technology has opened the door for lower cost and lower risk ways to study pedestrians’ behavior, perception of safety, and acceptance of safety technology while also offering a higher degree of data resolution and level or realism compared to previous pedestrian virtual simulators. The research presented in this dissertation addresses the development of a VR simulator for studying pedestrian safety, a validation analysis of the immersive virtual environment against pedestrian behavior in the real-world environment, and a safety analysis of alternative technology treatments at the uncontrolled crossing to prove the efficacy of using VR technology without the risks, time, and costs of real-world studies and safety analyses. 

Comparisons between real world and VR pedestrian behavior showed no statistical differences in gap acceptance through the use of chi-squared analysis and crossing speed through the use of independent samples t-test at a confidence level 95%. 94% of subjects felt that they were immersed in the virtual environment and 86% felt that their experience in the virtual environment was consistent with their real-world experiences. The results from this analysis prove that the use this VR simulator is a valid approach for studying pedestrian safety at uncontrolled crossings.

Safety analysis of the unsignalized crossing within the VR environment showed beneficial correlations when incorporating alternative safety technologies through bivariate correlations. Pedestrians were able to cross the street at slower, safer speeds, rather than darting out in front of approaching vehicles, regardless of the gap size between vehicles because they were able to communicate their intent to cross with approaching vehicles. 56% of subjects reported that they felt safe crossing the road using the mobile phone application, whereas 90% of subjects felt safe crossing the road with the flashing beacons. Compared to the 26.5% of subjects who reported that they felt safe crossing the street without alternative technologies, it can be concluded that the crossing alternatives increase pedestrian safety, both behaviorally and perceptively, at the crossing.

PHD (Doctor of Philosophy)
pedestrian, virtual reality, safety, connected vehicle, simulator
All rights reserved (no additional license for public reuse)
Issued Date: