The Insurance Institute for Highway Safety (IIHS) awarded the 2021–22 Tesla Model Y its top award, Top Safety Pick+, because it meets all the criteria for the “plus” designation. The IIHS noted that for a vehicle to qualify for either of its two awards, it must have earned good ratings in six IIHS crashworthiness evaluations. These include the driver-side small overlap front, passenger-side small overlap front, moderate overlap front, original side, roof strength, and head restraint tests.
The IIHS stated:
“Model Y vehicles built after April 2021 meet all the criteria for the ‘plus.’ Following a conversion to a camera only system, the standard front crash prevention system earns superior ratings in both the vehicle-to-vehicle and vehicle-to-pedestrian tests. The two available headlight systems earn good and acceptable ratings.
“Separately, the standard front crash prevention system on 2021–22 Tesla Model 3 vehicles built after April 2021 also earns a superior rating in both crash avoidance tests, following a software update. Vehicles built earlier earn an advanced rating in the vehicle-to-pedestrian evaluation.”
In April 2021, Tesla announced that it was doing away with radar on its new vehicles and going vision-only. Tesla and Elon Musk were both heavily criticized for this. However, the IIHS credited this switch to what earned its superior ratings in both vehicle-to-vehicle and vehicle-to-pedestrian tests.
I’d written an article in April that analyzed a video by Dr. Know It All who shared a study from Stanford that demonstrated how a Tesla vehicle can see through the fog. Although that article is technical, as is his video, Dr. Know It All showed how an AI system can be trained to see and break when it sees something unexpected in its path.
Although the technological advances in AI at Tesla’s AI Day event were in competition with the AI Bot that Elon unveiled (in terms of the media’s attention), there was a lot of detailed information as to how Tesla is improving its vision and AI in general.
During AI Day, Tesla’s Director of AI, Andrej Karpathy, explained the vision component that Tesla is designing. It’s a neural network that processes raw information from the vehicle’s 8 external cameras. Think of your eyes, for example, but Tesla has 8 positioned around the vehicle, and they don’t blink.
Andrej explained how the three-dimensional spaces of lines, edges, curbs, and other things you see on the road are turned into a “vector space” fed into Tesla’s AI system to determine where to drive. These include the positions and orientations of the vehicles and how far or close they are in proximity to you. Andrej showed videos of these raw images coming into the stack, being processed into the vector space, and how they then rendered on the display in the car.
AI and technical talk aside, the IIHS found that the Tesla Model Y has performed at the top safety level with this vision-based camera system, and this is a good thing for Tesla and its engineers who are working hard at making 100% full self-driving a reality.
Naturally, all of the engineering previously put into having Tesla cars protect their drivers and passengers in the event of a crash was also critical to the top score. For more on that, see our past coverage:
Appreciate CleanTechnica’s originality? Consider becoming a CleanTechnica Member, Supporter, Technician, or Ambassador — or a patron on Patreon.