Ghost driver: A field study investigating the interaction between pedestrians and driverless vehicles | IEEE Conference Publication | IEEE Xplore

How will pedestrians and bicyclists interact with autonomous vehicles when there is no human driver? In this paper, we outline a novel method for performing observational field experiments to investigate interactions with driverless cars. We provide a proof-of-concept study (N=67), conducted at a crosswalk and a traffic circle, which applies this method. In the study, participants encountered a vehicle that appeared to have no driver, but which in fact was driven by a human confederate hidden inside. We constructed a car seat costume to conceal the driver, who was specially trained to emulate an autonomous system. Data included video recordings and participant responses to post-interaction questionnaires. Pedestrians who encountered the car reported that they saw no driver, yet they managed interactions smoothly, except when the car misbehaved by moving into the crosswalk just as they were about to cross. This method is the first of its kind, and we believe that it contributes a valuable technique for safely acquiring empirical data and insights about driverless vehicle interactions. These insights can then be used to design vehicle behaviors well in advance of the broad deployment of autonomous technology.

How will pedestrians and bicyclists interact with autonomous vehicles when there is no human driver? In this paper, we outline a novel method for performing observational field experiments to investigate interactions with driverless cars. We provide a proof-of-concept study (N=67), conducted at a crosswalk and a traffic circle, which applies this method. In the study, participants encountered a vehicle that appeared to have no driver, but which in fact was driven by a human confederate hidden inside. We constructed a car seat costume to conceal the driver, who was specially trained to emulate

related reading

• Death of Elaine Herzberg - Wikipediaen.wikipedia.org
• Human-compatible driving partners through data-regularized self-play reinforcement learningarxiv.org
• Invisible Details of Interaction Designrauno.me
• Seeing Like a Sedan—Asteriskasteriskmag.com
• Building reliable sim driving agents by scaling self-playarxiv.org
• Moving Spaces—The Affective Embodied Self in Tram Design and the Autonomous Imaginary | Springer Nature Linklink.springer.com
• Interaction Models: A Scalable Approach to Human-AI Collaboration - Thinking Machines Labthinkingmachines.ai
• Connected and automated vehiclestc.canada.ca
• [2605.22748] Superhuman Safe and Agile Racing through Multi-Agent Reinforcement Learningarxiv.org
• Steering Behaviors For Autonomous Charactersred3d.com
• Researchclear.csail.mit.edu
• Self-driving police patrol vehicles tested on public roads in capital - Chinadaily.com.cnchinadaily.com.cn