Collaboration on the edge of chaos

Abstract

Research at the Institute for Simulation and Training has uncovered the curious fact that human psychomotor activity is mathematically chaotic at high performance levels. This chaotic behavior manifests both when humans are acting alone and when they are interacting with semi-autonomous devices in real and simulated environments. Other studies have reported that robots alone also exhibit mathematically chaotic behavior. This has led to the working hypothesis that chaotic measures such as the Lyapunov exponent can be used to quantify performance levels in human robot collaboration in an objective way. Experiments are in progress to help better understand and quantify the occurrence of chaotic behavior in human robot collaboration. The expectation is that Lyapunov exponent and other chaos-related measures will prove useful as diagnostic tools for both operational tasks and for training. Our goal is to investigate using overt actions (navigational inputs now and possibly other things like facial expression in the future) for capturing the Lyapunov exponent in real time and as a function that varies over time in response to behaviors. The challenge is to determine which factor to measure for a specific type of task. Additional research will be needed to link task to human psychophysical activity and to robot activity as well as the ability to transition between data modes as tasks change.