LCSR Seminar: Brent Gillespie “Predicting Human Behavior in Predictable Environments Using the Internal Model Principle”
All models are wrong, and too many are directed inward. The Internal Model Principle of control engineering directs our attention (and modeling proficiency) to what makes the world around us patterned and predictable. It says that driving a model of that patterned or predictable behavior in a feedback loop is the only way to achieve perfect tracking or disturbance rejection. In the spirit of “some models are useful”, I will present a control system model of humans tracking moving targets on a screen using a mouse and cursor. Simple analyses reveal this controller’s robustness to visual blanking and experiments (even experiments conducted remotely during the pandemic) provide ample support. Extensions that combine feedforward and feedback control complete the picture and complement existing literature in human motor behavior, most of which is focused on modeling the system under control rather than the environment.
Brent Gillespie is a Professor of Mechanical Engineering and Robotics at the University of Michigan. He received a Bachelor of Science in Mechanical Engineering from the University of California Davis in 1986, a Master of Music from the San Francisco Conservatory of Music in 1989, and a Ph.D. in Mechanical Engineering from Stanford University in 1996. His research interests include haptic interface, human motor behavior, haptic shared control, and robot-assisted rehabilitation after neurological injury. Prof. Gillespie’s awards include the Popular Science Invention Award (2016), the University of Michigan Provost’s Teaching Innovation Prize (2012), and the Presidential Early Career Award for Scientists and Engineers (2001).