Details of the incident

The video, which can now be found online, shows footage from the Tesla Model 3’s dashcam before it runs over a deer standing in the middle of the road. It was also clear that the car wasn’t trying to brake, Jalopnik writes.

The author of the video, who goes by the name Paul S on the social network At that moment the car was in fully autonomous driving mode. Even more surprising, the driver assistance system software did not stop even after the crash. Instead, traffic continued as if nothing had happened.

Big surprise after a dozen wrong stops every day,
– Written by Paul S.

The impact caused a large dent on the car’s hood, scratches on the windshield, and possibly more serious damage to the frame. “I’m experiencing the worst part of Tesla service right now. I can’t even get an estimate for insurance. Next meetings are only in January,” says the author of the video.

“FSD works great”

Can we expect the driver to change his mind after this? Certainly. However, this is not the case. As is often the case with die-hard Tesla fans, no problems, malfunctions or accidents lead them to suspect that the company’s technology is flawed.

Paul S immediately begins to justify FSD below the video:

Note: Although the deer was faintly visible in the video in the seconds before impact, it looked like just another bumpy road to the human eye. The post attracted some “wrong” attention from FSD deniers. FSD works great! This is just an extreme case. Maybe we need to teach FSD what a deer butt looks like
– writes more.

But in reality, this is not an “extreme case.” Deer collisions are common, with an estimated 2.1 million vehicle-animal collisions occurring each year in the United States. And if during normal driving the driver can still turn at the last moment, then if the autopilot does not learn to recognize animals, the more widespread its use becomes, the worse these statistics will only get.

Simply put, mainstream driving software – especially if it is called “Totally Self-Driving” – needs to be equipped with features to reliably handle such events.

Why does this problem exist?

Although we cannot know what the FSD system’s AI was “thinking” at the time, the possible cause of such events Lack of lidar technology in Tesla cars.

LiDARs are laser devices that emit invisible signals into the surrounding area and then record their reflections from surfaces. In this way, an environmental map is created in advance and obstacles can be detected even in pitch darkness.

For some reason, Elon Musk did not like this technology very much and therefore decided to leave it in the company’s cars. he proposes relies entirely on a camera system that visually detects the environment. Therefore, poor lighting, poor viewing angle, fog or other visual obstructions can cause similar phenomena where computer vision cannot recognize the animal or person in front.

This is why Tesla’s competitors use LiDAR, and some combine both lidars and cameras to provide greater confidence.