Imperfect Automation in Scheduling Operator Attention on Control of Multi-Robots

Abstract

An operator’s workload increases substantially when the operator must control multiple robots and continually shift attention from robot to robot. As the number of robots increases, the amount of time an operator can spend operating any particular robot decreases, which leads to inevitable changes in the robot’s performance. If the robots could self-report encountered faults, the operator could conserve cognitive resources to spend on reasoning about more complex situations. In the reported experiment, participants performed foraging tasks while assisted by an alarmed system, either Open-queue in which all alarms are displayed or SJF-queue (shortest-job-first), whose reliability level was high (90%) or low (50%) under different task load (3-robots vs. 6-robots). The results showed that simply increasing the system reliability might not effectively contribute to the overall performance or the participants’ trust in automation. An inverse relationship was observed between experienced workload and rated trust which also amplified the effects of imperfect automation.