Investigators

Siby Samuel
University of Massachusetts - Amherst
Industrial Engineering
Eleni Christofa
University of Massachusetts - Amherst
Civil Engineering
Michael A. Knodler, Jr., Ph.D
University of Massachusetts - Amherst
Civil Engineering

Search

Project

To Trust or Not to Trust? A Simulation-based Experimental Paradigm

Human error has been cited as a causal factor in 94 percent of vehicle crashes (Singh 2015). Automated vehicle technology promises to help decrease that share by at least partial takeover of vehicle control from drivers in certain conditions. This is expected to lead to an increase in traffic safety and performance. However, automation technology will not be as effective if drivers do not accept the technology and/or do not utilize it appropriately (Lee and See 2004). Recent incident demonstrate how over-trust (misuse) can negatively impact safety. A common thread throughout the literature suggest that an appropriate level of trust that matches system capabilities is necessary to gain full benefit from that system. Failure to gain the appropriate level of trust on automation, whether it is over trust (mistrust) or under-trust (distrust), would diminish the positive effects of that technology on safety and performance of driving (Riley 1989, Parasuraman et al., 1997; Parasuraman and Riley 1997; Lee and See 2004; Visser et al., 2014). The current literature on driver’s trust in automated vehicles shows a paucity of research to examine how well drivers trust in automated vehicles matches the trustworthiness of the system. Furthermore, there is a specific need to how drivers’ trust on the system changes based upon experiences within the system, including failures in systems performance.

This research is going to address these gaps and study how drivers’ levels of trust change when they interact with systems under varied levels of trustworthiness. This study will employ a full-scale driving simulator, a head-mounted eye tracker, and a set of questionnaires (before starting the experiment, midway through automation experiment, and after they completed the experiments) to examine drivers’ interaction with automated vehicles. Each subject driver in this experiment is assigned to an automated vehicle with specific level of reliability. Subjects driving performance, as well as glance behavior, willingness to engage in the driving task during the experiment, and questionnaires data will be used to measure their subjective and objective level of trust and the sensitivity of their trust after encountering failures in the system