Exploring the Development of Trust Between Human and Autonomous Teammates
Abstract
This research aimed to identify and combine objective performance-based metrics of trust with subjective measures in order to better understand the development of trust between humans and autonomous systems during complex tasks that include risk. A virtual reality game developed by the Johns Hopkins University Applied Physics Laboratory (JHU-APL) was the platform used to measure trust scores between both homogeneous human teams and human-autonomous teams (utilizing a bot teammate designed by JHUAPL). Of interest in this study was whether objective and subjective measures of trust differed depending on whether the teammate was (a) actually a human or a bot and (b) perceived to be a human or a bot. Additionally, this study examined how objective performance metrics compare to subjective trust scores among varying teammate conditions. The objective performance metrics identified in this thesis were not indicative of overall trust alone but did shed some light on the development of trust and would be beneficial if used in conjunction with subjectively derived metrics to provide a more complete set of measures that can contribute to a greater understanding of the development of trust between man and machine.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2020
- Accession Number
- AD1126390
Entities
People
- Peter J Coombe
Organizations
- Naval Postgraduate School