Traditional approaches to autonomous vehicles (AVs) rely on using millions of miles of driving data in conjunction with even more miles of simulated data as inputs to supervised machine learning ...