State of the art automated vehicles can safely navigate planned paths which do not exceed limitations imposed by friction and other physical properties. Planning trustworthy paths, however, is challenging when precise limitations are unknown. Emergency maneuvers are especially likely to require an automated vehicle to use the limits of friction to escape an accident.
My research focuses on contingency planning for situations when limit behavior is desired, but those limits are also unknown. To accomplish this, I employ a model predictive control framework which considers several potential paths simultaneously.