Video of a presentation explaining how the Emergency Auto Brake System, one of the safety features of the Autopilot driving experience in Tesla’s vehicles, stopped Tesla vehicles before they hit the pedestrians.
Autopilot, the semi-autonomous driving support found in Tesla’s vehicles, is among the most spoken features of the cars. Autopilot offers some safety features to its users as well as autonomous driving features.
Emergency Auto Brake System is among the important safety features of Autopilot. Emergency Auto Brake, which enables Tesla vehicles to stop in unexpected situations, saves the lives of careless pedestrians.
Tesla announces the operating logic of the Emergency Auto Brake system
In Euro NCAP held last year, we saw how and how successful this Autopilot supported safety feature was in the tests against pedestrians. Tesla shared real-world Emergency Auto Brake moments this time after last year’s tests.
In his presentation at the Scaled Machine Learning Conference, Andrej Karpathy, the head of Tesla’s artificial intelligence and computer vision, shared videos about how Tesla cars saved pedestrians’ lives thanks to the automatic braking system. The images of the conference, which took place at the end of February, were newly shared online.
In three video examples shown by Andrej Karpathy at the conference, it is seen that the pedestrians stopped just in time with the pedestrians entering the vehicle’s field of vision. Tesla can capture and record these stopping moments of its vehicles thanks to the TeslaCam feature in the vehicles.
In his presentation, Karpathy explained that the cameras of Tesla vehicles continue to work even if the vehicle is not even in Autopilot, and stops when a pedestrian is in front of it. Karpathy said there are other videos of examples where Tesla vehicles stopped without stopping pedestrians with the Emergency Auto Brake.
While Tesla’s head of artificial intelligence and computer vision stated that Tesla continued to work on systems that will provide full autonomous driving, he explained how they used machine learning in this process.
Andrej Karpathy’s presentation at Scaled Machine Learning Conference: