Development of an autonomous manager for dyadic human-machine teams in an applied multitasking surveillance environment

Abstract

Automation is crucial in increasingly many workplaces. Though automation is often associated with job replacement, humans and machines have divergent proficiencies. Thus, human-machine teaming is generally favored over replacement. Within applied surveillance environments, automation is leveraged for cognitively intensive tasks. To maintain optimal performance within a dyadic human-machine team, we developed an Autonomous Manager (AM) that dynamically redistributes tasks between human and machine. Participants performed four simultaneous image identification tasks while paired with a simulated autonomous partner. Our AM was responsible for monitoring team performance and redistributing tasks when performance fell sub-threshold. We manipulated the refresh rate of the images, affording us the opportunity to measure improvement under multiple conditions.