An Experimental Investigation of how Robotic Learning and Cloud-Based Information Affects Trust in Human-Machine Teaming Contexts

In this proposed research project, we will seek to understand the effects of robot learning and changes in access to cloud-­-based systems on the predictability of human machine teaming (HMT) systems, and in turn, on a human operator?s trust toward their robotic teammates. The three-­-year research plan has three interrelated and mutually supportive experimental objectives: Experiment 1 assesses how a human teammate?s trust calibration is affected by unpredictable robotic behavior; Experiment 2 investigates how this unpredictability is affected by the type of the machine?s learning algorithms and the degree of access to cloud-­-based systems; and Experiment 3 determines how transparency, in the forms of robot-­-to-­-human and robot-­-of-­-human, can be used to improve trust calibration. We will use this understanding to help inform guidelines for the design of HMT systems (via system transparency, operator interfaces, and operator training techniques) for use in multi-­-robot and human teaming environment. We will use a recently awarded instrumentation grant from the Air Force Office of Scientific Research (AFOSR) Defense University Research Instrumentation Program (DURIP) to build a test bed environment to support this research. The research will be conducted jointly with Dr. Joseph Lyons, who is a Technical Advisor for the Human Trust and Interaction Branch within the 711 Human Performance Wing at Wright-­-Patterson Air Force Research Lab. In addition, a project advisory board consists of three subject matter experts who are currently active in conducing human machine teaming research will be formed to provide technical feedback and assist the project team.