Modeling Operator Performance in Low Task Load Supervisory Domains
[Technical Report, Master's Thesis]
Massachusetts Institute of Technology
Pagination or Media Count:
Currently, numerous automated systems need constant monitoring but require little to no operator interaction for prolonged periods, such as unmanned aerial systems, nuclear power plants, and airtraffic management systems. This combination can potentially lower operators workload to dangerously low levels, causing boredom, lack of vigilance, fatigue, and performance decrements. As more systems are automated and placed under human supervision, this problem will become more prevalent in the future. To mitigate the problem through predicting operator performance in low taskload supervisory domains, a queuing-based discrete event simulation model has been developed.To test the validity and robustness of this model, a testbed for single operator decentralized control of unmanned vehicles was utilized, simulating a low workload human supervisory control HSC environment. Using this testbed, operators engaged in a four-hour mission to search, track, and destroy simulated targets. Also, a design intervention in the form of cyclical auditory alerts was implemented to help operators sustain directed attention during low task load environments.The results indicate that the model is able to accurately predict operators workload. Also, the model predicts operators performance reasonably well. However, the inability of the model to account for operator error is a limiting factor that lowers models accuracy. The results also show that the design intervention is not useful for operators who do not have difficulties sustaining attention for prolonged periods. The participants of this study were exceptional performers, since most of them had very high performance scores. Further research will investigate the possibility of conducting another low task load, long duration study with a more diverse set of participants to assess the impact of the design intervention and to extract personality traits that may affect system performance.
- Personnel Management and Labor Relations