Using Wireless Communication to Control Road-user Interactions in the Real World

Paper in proceeding, 2017

Autonomous vehicles promise to conquer the road network, revolutionize transportation, and improve traffic safety in the near future. However, before they can take over, they will be sharing the infrastructure with manned vehicles, and will be expected to behave like those vehicles do in traffic. In other words, the first autonomous vehicles will have to remain within human comfort boundaries so as not to surprise, confuse, or scare any road users. For safety reasons, comfort boundaries in critical situations can only be measured in driving simulators or on test tracks. In less critical situations, field and naturalistic studies provide the most realistic estimations of comfort boundaries. In fact, naturalistic data is collected in the real world by drivers following their daily routines; thus they capture realistic driver behavior. However, when we want to assess comfort boundaries in specific scenarios — such as intersections — with complex interactions, even large databases may offer limited data with great variability. This study presents and verifies a new experimental methodology to diminish variability in naturalistic and field data collection without compromising ecological validity. This methodology controls the environment in real time, depending on the behavior of study participants to produce specific driving situations in the real world. Recreating similar driving conditions over and over increases data consistency, enabling data to be averaged across participants and repetitions. This paper estimates comfort-zone boundaries at intersections from naturalistic data. Potential applications for this methodology include the development and evaluation of advanced driving assistance systems, as well as the design of test procedures for active safety and cooperative systems.