Situational Cues for Continuous Trust Calibration in Automated Systems

Date

7/10/2022

Journal Title

Journal ISSN

Volume Title

Publisher

51st International Conference on Environmental Systems

Abstract

Appropriate user trust calibration in automated systems is critical for optimizing system usage, improving task performance, and reducing user workload. While undertrust in a system may lead to system disuse and suboptimal performance, overtrust in a system can result in reduced user situation awareness and susceptibility to consequences of system failure. In dynamic domains, the reliability of an automated system may fluctuate based on environmental conditions and the type of task being performed. Fluctuation in system reliability demands that user trust in the system adapts to optimize system usage and task performance. Recent attention has been focused on the presentation of adaptive trust calibration cues based on quantified human operator trust to promote appropriate trust calibration. In dynamic domains such as future human and robotic space exploration missions, real-time quantification of operator trust in an automated system may not be possible. In this work, we expand the application space of trust calibration cues through the introduction of situational trust cues (STCs), presented independent of user real-time trust in the system. Situational trust calibration cues are presented if environmental conditions or task type is changed in a manner that significantly affects performance of the automated system. We present the design of an experiment investigating the application of STCs to a Lunar rover path planning automated decision support tool in a dynamic mission operations environment. The experiment is designed to assess the impact of STCs on user selection of the most appropriate level of automation to afford an automated decision support tool, where the most appropriate level of automation fluctuates between three automation levels based on the type of planning task being executed and the state of the simulated environment. We ultimately plan to investigate the utility of STCs in mitigating user overtrust and undertrust in automated systems.

Description

Alexandra Forsey-Smerek, Massachusetts Institute of Technology, US
Katya Arquilla, Massachusetts Institute of Technology, US
Julie Shah, Massachusetts Institute of Technology, US
ICES405: Human/Robotics System Integration
The 51st International Conference on Environmental Systems was held in Saint Paul, Minnesota, US, on 10 July 2022 through 14 July 2022.

Keywords

trust calibration, trust cues, decision support tools, automated systems, human-automation collaboration

Citation