Tesla He has published a series of videos showing how the Autopilot emergency braking system of his vehicles responds to the surprise appearance of pedestrians crossing the street.
The examples were shown during a presentation made by Andrej Karpathy, director of artificial intelligence and artificial vision of the company during the conference ScaledML 2020 It was held in late February at the Historical Museum of Computers in Mountain View, in the heart of Silicon Valley.
In the presentation, Karpathy explains that they have approximately a million vehicles circulating on the streets around the world and constantly training the artificial intelligence of the Tesla Autopilot and shows some examples of how it is capable of reacting and saving lives in real situations that occurred and were recorded by car cameras.
From minute 3:30 You can see three different situations of the emergency braking activation. In the first example it is a person who appears out of nowhere trying to cross a street when the Tesla is already in motion. In the second, the driver does not detect the presence of a pedestrian and the car reacts. In the last example, you can see how other vehicles prevent you from seeing the pedestrian crossing but the emergency system is activated and prevents the accident.
Tesla activates emergency systems to avoid collisions hundreds of times a day throughout its fleet
Karpathy explains that Tesla go hundreds of similar cases every day and that in all these cases the emergency reaction system has been able to avoid major mishaps. It also emphasizes that in all three cases, the vehicles were being driven and the Autopilot was not connected, but that collision avoidance systems are always activated in extreme cases like these.
What follows in the presentation is an extremely interesting explanation of how Tesla is looking to use artificial intelligence to detect depth in images recorded by vehicle cameras and sensors, and create a virtual version of the LiDAR sensors.
Is the Tesla Autopilot really capable of saving lives on the road?
👇 More in Hypertextual