Improved SA of Automation Functioning Promotes Appropriate Automation Usage
Across the world, automated systems are being developed for use in many environments and industries to increase productivity, safety, and reliability. Even the most advanced automated systems, however, generally require human interaction, which may include monitoring, selecting from options generated, or action authorization. When these systems fail to produce the expected benefits, the fault often lies not in the automation, but rather in the human-automation collaboration. Situation awareness (SA) is widely recognized as a critical precursor to decision-making, and operator SA on automation functioning is necessary to support effective human-automation teaming and appropriate automation usage decisions (AUDs). Many investigations of AUDs have focused on the development of trust in automation, but many of these studies have focused on random automation failures, that is, failures that occur without a clear rationale which “makes sense to the human operator. In real world operations, automation often fails for reasons that are rational if the operator understands the automation functional parameters and limitations. A goal of the current research is to speed the development of trust by promoting an understanding of automation functioning that includes automation limitations that are likely to produce failures. We present the results of two experiments in a dynamic, fast-paced simulation environment where study participants must allocate units to objects with the support of agents that make tasking recommendations. For each recommendation, participants can accept or reject all or part of the agents unit allocations. In Experiment 1, the “systematic errors of the agents were caused by their failure to include a critical assessment into tasking recommendations. In a knowledge manipulation, participants in the experimental group were informed of the source of automation unreliability (i.e., the agent does not consider a critical factor) while participants in the control group were aware of the significance of the factor without knowing that it was the source of agent error. Findings from this study indicated that understanding the source of automation error facilitated appropriate automation usage decisions and improved SA on the factors that contributed to automation errors. In Experiment 2, we investigated AUDs under conditions of predictable failure versus random failure. Overall, study results showed that participants in the systematic error condition made more appropriate automation usage decisions. The paper concludes with important “lessons learned and directions for future research.
Laura D. Strater John Yen Michael McNeese Haydee M. Cuevas Tim Hanratty
SA Technologies College of Information Sciences and Technology, The Pennsylvania State University The U.S. Army Research Laboratory
国际会议
17th World Congress on Ergonomics(第十七届国际人类工效学大会)
北京
英文
1-9
2009-08-09（万方平台首次上网日期，不代表论文的发表时间）