Modeling Trust in Heterogeneous R3 Human-Machine Teams

Lead Researchers 

• Kevin Zemlicka, Academic Advisement, Anthropology & Psychology
• Nhut Ho – ARCS Director – Mechanical Engineering

Collaborators

• Olivier Toupet, Jet Propulsion Laboratory
• Nelson Brown, NASA Armstrong Flight Research Facility
• Ali Agha, Jet Propulsion Laboratory

Student Team

• Julia Spencer, Tran Le
• Jessica Steiner
• Dana Bellinger
• Samuel Mercado
• Andranik Yelanyan*
• Shaun Cabusas*
• Karen Dominguez*
• Cesar Flores*

Funding

• Funding Organization: Air Force Office of Scientific Research
• Funding Program:

Abstract

The current project has identified a gap in research studying trust evolution and calibration in R3 contexts, that is, in a heterogenous team of multiple real highly-autonomous machine teammates and real human teammates operating in situations with real consequences. To address this gap, the current 5-year study seeks to obtain foundational lessons and insights on how trust is calibrated and evolves over time, identify how technology and non-technology-related factors (e.g., organizational, cultural, personal) influence trust evolution, validate extant theoretical trust models against trust calibration and evolution of heterogeneous R3 human-machine teams (HMTs), generate hypotheses for trust evolution and calibration in heterogeneous R3 HMT contexts, and formulate research questions and identify pathways to translate the results of this basic research project to applied research. The field study will be designed in the form of two case studies, one for each heterogeneous R3 HMT: NASA JPL’s Mars 2020 HMT and NASA JPL’s Team CoSTAR subTerranean robotics HMT. The researchers will employ a set of complementary qualitative methods: case study/extended case study, iterative Rapid Assessment Procedures, and a grounded theory approach. Potential contributions to the DoD and Air Force’s missions include recommendations for building appropriate trust in heterogeneous R3 HMTs, basic research results transitioned to applied research for current Air Force Research Laboratory initiatives, and informing advanced concepts of operations that involve heterogeneous R3 HMTs and related advanced autonomy technologies envisioned in the DoD autonomy roadmap.

There has been an increased interest in human machine team trust within the past few decades, though much of existing research consists of laboratory-based or simulation-based environments. While such studies are valuable in understanding solution approaches, there is a current lack of literature describing trust evolution and calibration in real R3 contexts (involving real heterogeneous human-machine teammates, real systems, and real consequences). R3 systems are a powerful context for trust research because they provide high levels of face validity and offer real-life richness from which both technology and non-technology factors that influence trust can be identified. Beyond the lack of trust literature in R3 contexts, another important limitation of previous HMT trust literature is that most studies have been confined to a single operator interacting with a single machine. However, existing and proposed HMT architectures take on multiple forms allowing for complex heterogeneous teaming, e.g. multi-robot technologies categorized into team agents, hierarchical agents, and mixed initiative systems. The current project seeks to fill this gap through the proposal of a 5-year study, employing qualitative methodologies to inductively explore heterogeneous R3 HMTs in unique, state of the art active NASA projects.