Human performance on model reference adaptive control systems with human reaction delay

Model Reference Adaptive Control (MRAC) is capable of dealing with system uncertainties effectively when controlling real-world systems. Such a capability can be beneficial in human-in-the-loop robotic control settings, where human and robot uncertainties can be handled properly to render the human-machine interface stable, and potentially highperforming. In this thesis, an MRAC architecture is designed and experimented with volunteering human subjects, with the guidance of recent theoretical results considering human reaction delays. To this end, stabilizing designs based on linear quadratic regulators (LQR), modified LQR, and pole placement are pursued, and the stability of the closed-loop system is then analyzed with respect to human reaction delays, using TRACE-DDE toolbox assuming that the human model is given by a Neal-Smith pilot model. The control design efforts help determine the appropriate settings for studies in experimental settings. Here, we develop three design conditions for MRAC to experiment with human subjects, under approved NU IRB protocol. Specifically, we designed a Cursor Control Game for the subjects to play, where a subject uses the computer mouse to move a cursor on the screen for tracking purposes. While performing this task, an MRAC in the background assists the subject's mouse commands in order to help the subject in the tracking game. According to the data collected in the experiments with human subjects, performance indexes are defined and three different designs of MRAC systems are compared based on both human performance and human control effort. The comparison shows that the faster