Yes its good but the data is used to show that "it is safer" however we do not have the unsupervised data to tell if the system is safer or not. What we can say is the supervised version is safer vs human driver alone however not necessarily safer than unsupervised which is the goal
@teslaeurope Nobody ever points out that this is mainly because the software is doing it supervised. ...... So the car and driver are being vigilant vs the car alone. Im sure that would slew the stats