George Gunter is a PhD candidate in the Civil Engineering department at Vanderbilt University, and a member of the Institute for Software Integrated systems. George's work focuses on safety and security in applications of CPS to civil infrastructure. This has included research into cyber-security of mixed-autonomy traffic flows, as well as the use of formally verified control theoretic approaches to create safety guarantees for automated vehicles.
This work looks at experimental testing of the use of a control barrier function (CBF) for safe control of an automated vehicle (AV). An experimental dataset is collected by running a CBF on a live AV that assumes an instantaneous acceleration model, and failure to satisfy a safety property is found. On the collected data, a system identification technique is run that finds a more accurate model which assumes dynamical lag on acceleration. It is shown that the CBF induced by this new model subsequently induces a need to estimate accelerations of other vehicles. The tradeoff between error due to modeling and error from sensors/estimation is given.