The Effects of Level of Autonomy on Human-Agent Teaming for Multi-Robot Control and Local Security Maintenance

摘要

Abstract : The effect of differing levels of autonomy (LOA) using a route-planning agent (RoboLeader) on human multitasking performance (convoy navigation management, target detection, and maintaining situation awareness) was evaluated using a simulated environment. Eye tracker data was used to further explain performance findings, as well as examine Operator Control Unit usage among participants. Increasing LOA improved task performance and reduced workload; however differential effects due to spatial ability (SpA) were found that indicate an optimal level of LOA. Although low SpA participants performance improved with increasing LOA, high SpA participants had a performance decline in the highest LOA. Participants with high perceived attentional control (PAC) benefited from increasing LOA, while those with low PAC did not. Low SpA participants SA improved in the RoboLeader conditions to nearly that of high SpA participants.