Impact of Automation's Level and Reliability on Multi-Vehicle Supervisory Control Task Performance

Abstract

Supervisory control of multiple unmanned vehicles raises many questions concerning the balance of system autonomy with human interaction for optimal operator situation awareness and system performance. In the research reported here, a previously used experimental protocol was extended such that both automation level and reliability of two tasks were systematically manipulated, rather than just a single task. The autonomy level across the two tasks was either different (i.e., one task was more automated than the other) or similar for two primary control tasks completed in tandem: allocation (assignment of sensor tasks to vehicles) and router (determining vehicles’ flight plans). Additionally, the automation’s reliability for both tasks was systematically manipulated. Objective data were collected from twelve participants on these tasks as well as multiple secondary tasks in a multiple autonomous vehicle simulation. The results provided further evidence that the autonomy level of one task can influence the performance of other tasks and that finegrained variation of autonomy level across tasks may negatively affect an operator’s awareness of the automation’s current configuration.