Cognitive Models Predicting Surprise in Robot Operators


Robots and their remote human operators do not always form a cohesive team. Where robots make autonomous decisions, operators can be surprised and will, consequently, lose trust in automation and end up micro-controlling the robot. The benefits of partial autonomy are lost in the process. We ask whether a system using a cognitive model can predict operator surprise, pre-empt it with verbal explanations, and restore trust (where appropriate) and task performance among operators.