Citation

Manser MP, Campbell J, Fincannon TD, Krake AK, Hoekstra-Atwood L, Crump CE, Wu LT. 27th International Technical Conference on the Enhanced Safety of Vehicles (ESV); April 3-6, 2023; Abstract #: 23-0342, pp. 13p. Washington, DC USA: US National Highway Traffic Safety Administration, 2023 open access.

Copyright

(Copyright © 2023 open access, US National Highway Traffic Safety Administration)

Abstract

27th International Technical Conference on the Enhanced Safety of Vehicles (ESV): Enhanced and Equitable Vehicle Safety for All: Toward the Next 50 Years

https://www-esv.nhtsa.dot.gov/Proceedings/27/27ESV-000342.pdf

Research Question/Objective: In transportation, mental models are essential to mobility and safety because drivers rely on them to understand how to interact properly with their vehicles, the transportation infrastructure, and the environment. Poor performance and errors can occur when a driver acts in accordance with inaccurate mental models. Mismatches between mental models and actual experiences can also lead to reduced trust when, for example, the system with which they interact fails to perform to their expectations. The current study examined differing information types regarding Automated Driving System (ADS) capabilities and limitations on development of mental models and trust while using simulated Level 3 (L3) systems and a "dual model" use case of Level 4 (L4) systems (i.e., the vehicle can be both manually operated and can be controlled by ADS in certain ODDs).

Method and Data Sources: 48 females and males between the ages of 25 and 65 had four exposures to L3 and L4 systems in a driving simulator. Participants used either a basic human machine interface (HMI) that indicated the ADS was active, or they used an enhanced HMI that provided additional information indicating when the system was experiencing limitations (e.g., regarding detection of degraded lane lines). Participants used a simulated Level 3 system for two exposures and a simulated Level 4 system for two exposures. The acquisition and development of mental models and trust were assessed with standardized questionnaires.

Results: Regardless of exposure to each system over time, participants' mental models were more accurate for the simulated Level 4 system compared to the simulated Level 3 system and trust was greater for the simulated Level 4 system during the second exposure.

Discussion and Limitations: This paper summarizes research an ongoing project, and a final report will be published at a later date. Results of the current work suggests that the acquisition and development of mental models and trust can be differentially impacted by how well the ADS performs and the level of automation. However, because the study relied on simulated Level 3 and Level 4 systems, the results may not represent real world implementations of the technology.