Analysis of Facial Expressions: Explaining Affective State and Trust-Based Decisions During Interaction with Automation
Abstract
Trust is a critical factor in the development and maintenance of effective human-autonomy teams. This becomes more important as the technology advances in independent and interdependent decision-making with humans, especially in high-risk dynamic environments. As such, new processes are needed to classify an individuals affective state change that could be related to either an accurate or a misaligned change in trust that occurs during collaboration. The task for the current study was a simulated leaderfollower driving task with two different types of driving automation (Level 2: full vs. Level 1: speed only) and across two different automation reliability levels (good vs. bad). Facial expression analysis and subjective questionnaire measurement were evaluated to gauge group differences in affect-based trust. Through a novel analysis approach, results indicated that the participant sample was best described by four distinct group clusters based on demographics, personality traits, response to uncertainty, and initial perceptions about trust, stress, and workload associated with interacting with the driving automation. These groups showed marked differences in their level of subjective trust and affect via facial expressivity. This suggests that trust calibration metrics may not be equally critical for all groups of people.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 28, 2020
- Accession Number
- AD1098113
Entities
People
- Brandon S. Perelman
- Catherine Neubauer
- Claire La Fleur
- Gregory Gremillion
- Jason S. Metcalfe
- Kristin E. Schaefer
Organizations
- United States Army Research Laboratory