Validation of Scanning and Complacency Models for Robotic Arm Control

We develop an automation complacency model of the human operator/supervisor of a robotic arm designed for space missions. Manual performance and two degrees of automation are modeled for arm trajectory control in a 3-segment scenario: a visual guidance support to manual control, and a full automatic control. The automation functions perfectly for several trials but then unexpectedly fails, in the same manner for both conditions. We model complacency-based visual scanning, and how that scanning will predict differences between conditions in the noticing of automation failures. Both scanning across the workspace and failure detection are validated with human-in-the-loop simulation data.