Human-centered time-pressured decision making in dynamic complex systems

Many real-world applications are complex, dynamic, and uncertain. Human operators play an important role in ensuring the safety and in achieving operational effectiveness in such systems. Humans work with computerized processes in monitoring the system state and are often involved in re-planning, troubleshooting, and supervisory control tasks. During task performance, both humans and computerized processes bring in varying strengths and limitations. Research on human-centered automation in aviation, satellite ground control, and nuclear power plant control has resulted in broad guidelines on system design involving human and computerized processes in supervisory control. However, problems such as increased human error, lack of situational awareness, and opacity from poorly automated systems remain, particularly in scenarios where human operators must make decisions in time-pressured planning. While anecdotal evidence does exist that interactive systems are better than completely manual or completely automated systems, there is a lack of systematic studies of human-centered modeling in joint cognitive systems. This research addresses the issue of joint cognitive problem solving for a class of problems related to supervisory control of vehicle routing. The key research question addressed by this study is whether a human integrated approach helps in better generation of alternatives and better evaluation of alternatives that would potentially lead to better solutions to problems. This research aims to develop a systematic approach to couple humans and machines to include effective task partitioning. The hypotheses related to the research questions are empirically evaluated, using human subjects, in prototypical realistic scenarios. Empirical results from a simulated military mission indicate that the human integrated approach resulted in better overall performance when compared to purely automated solutions for vehicle routing problems considered in this research study. Specifically, significantly more high priority targets were covered in the human integrated approach compared to the automated solution without any significant degradation with respect to all the other dependent measures including percentage of total targets covered, low priority targets covered, total targets covered in threat zone, high priority targets covered in threat zone, and low priority targets covered in threat zone. Results also indicated that the graphical representation of the alternatives leads to quicker evaluation time than tabular representation. One of the primary contributions of this research is a framework to demonstrate the rationale of using joint cognitive systems in time-pressured decision making. Some of the other contributions include a methodology for the evaluation of alternatives in a cognitively effective manner, and a baseline to compare results of future studies in interactive modeling.