Cross Platform Validation of Pedestrian Simulators – PEDSIVAL

While the development of safety measures in cars successfully reduced the accident rate overall, the fatality rate of pedestrians is still considerably high, especially in urban environments. In 2013, 22% of all killed persons in traffic accidents in the EU were pedestrians. Furthermore the fatality rate of pedestrians has decreased by only 11% since 2010 compared to the 18% decline of all traffic participants. The majority of pedestrian fatalities in the EU (69 %) arise in urban areas. (European Commission, 2015)

Whereas driving simulators have been already for decades a valuable tool for investigations of human behavior and validation of advanced driver assistance systems (ADAS) – regarding the driver – the development of pedestrian simulators is still in the fledgling stages. These kind of pedestrian simulators are used to investigate human behavior – regarding the pedestrian – in urban traffic scenarios in a reproducible, safe and cost efficient way. They enable studies on pedestrian behavior in hazardous traffic situations (e.g. crossing scenarios). Being an indispensable tool for car manufacturers and OEMs for the investigation of vehicle-pedestrian interactions in the development process of ADAS (pedestrian detection and avoidance), the importance will even grow in the course of the appearance of autonomous and silent cars (e.g. electric vehicle).

The French research institute IFSTTAR and the German university TU München have been in this field on the cutting edge from the very beginning. Independently, they developed two different types of pedestrian simulators using different technological approaches. In 2003, the IFSTTAR has been the first research institute in the world building a full-scale pedestrian simulator in which the participants can actually walk to cross streets within a CAVE (Cave Automatic Virtual Environment). In 2014, the TU München built an HMD (Head-Mounted Display) based pedestrian simulator thanks to the technological advances made in the field of virtual reality (VR). This device has the advantages of being cheaper and portable.

Whereas other existing pedestrian simulators use locomotive devices such as walk-in-place platforms (e.g. treadmills) or joysticks to simulate the participant’s move in virtual environments, the pedestrian simulators of the two project partners support natural walk. While the participants can physically walk in both simulators, the images and sounds in the two simulators show differences that could affect pedestrians’ behavior. This project aims at answering the question on the validity of the two simulators for pedestrian safety research.

By running highly comparable experiments on both simulators, the technological impact on pedestrian behavior will be investigated and a cross-platform validation performed. Furthermore, in order to assess absolute validity, both simulators will be compared to experiments in real life under similar conditions.