Enhancing Decision Making through Adaptive Trustworthiness Cues

ABSTRACT: This proposal is for an Adaptive Trustworthiness Calibration Interface (ATCI) using real-time measurement and cognitive agent modeling. Our objective is to enhance an operator's decision making by providing trust cues so that the perceived trustworthiness matches the actual trustworthiness of the system, thus yielding calibrated trust. There are many issues of trustworthiness in the military, but the most pressing today is trust in intelligent machine systems as the military moves to a more robotic battlefield. The most important current robotic systems are Unmanned Aerial Vehicles (UAVs), and in particular systems of multiple UAVs. We will provide the operator with visualization tools to diagnose the actual trustworthiness of the system by showing the risk and uncertainty of the associated information. We will measure and model the trust of the operator in real-time, and adapt the display so that an operator who wrongly trusts too much is encouraged to trust less and an operator who wrongly mistrusts is encouraged to trust more. We focus on both individual operator"s trust and the transparency of the system. Three innovative elements are central to our approach: Trustworthiness Diagnosis Displays, Quantitative Modeling of Trust Calibration in Cognitive Agents, and Adaptive Trust Cueing and Real-Time Interface Calibration. BENEFIT: Three factors converge to create the need for a military decision-maker to assess the trustworthiness of information in automated systems; these are: uncertainty in data fusion, trust in advanced decision support, and ubiquitous situational awareness. These conditions require a human-machine interface that will enhance an operator"s decision-making, identify uncertainty in underlying information, and calibrate trust adaptively to the operator. The complete ATCI will provide an infrastructure and method for control decisions, with guidance based on both the benefits and costs of allocating resources to further uncertainty reduction, and on the value of specific collection and analysis options in terms of their potential impact on decisions and UAV mission outcomes. To ensure the utility of our solution we will include in Phase I a functional Proof-of-Concept Demonstration based on a relevant and realistic use case involving a single operator coordinating, monitoring, and controlling multiple cross-platform UAVs. Emphasis will be given to the user interface and enhancement of the usability of the integrated system. To achieve this we will conduct human factors and usability analysis of the system functions and user interface. Our approach is supported by an on-going Air Force sponsored SBIR Phase II effort to develop an Adaptive Interface Management System (AIMS) for the USAF Vigilant Spirit Control Station (VSCS).