The U.S. government’s road safety agency is again investigating Tesla’s “Full Self-Driving” system, this time after getting reports of crashes in low-visibility conditions, including one that killed a pedestrian.
The National Highway Traffic Safety Administration says in documents that it opened the probe on Thursday with the company reporting four crashes after Teslas entered areas of low visibility, including sun glare, fog and airborne dust.
In addition to the pedestrian’s death, another crash involved an injury, the agency said.
Investigators will look into the ability of “Full Self-Driving” to “detect and respond appropriately to reduced roadway visibility conditions, and if so, the contributing circumstances for these crashes.”
This of course assumes 1) that cameras are just as good as eyes (they’re not) and 2) that the processing of visual data that the human brain does can be replicated by a machine, which seems highly dubious given that we only partially understand how humans process visual data to make decisions.
Finally, it assumes that the current rate of human-caused crashes is acceptable. Which it isn’t. We tolerate crashes because we can’t improve people without unrealistic expense. In an automated system, if a bit of additional hardware can significantly reduce crashes it’s irrational not to do it.
This is directly a result of Elon’s edict that Tesla cars don’t use lidar. If you aren’t aware Elon set that as a requirement at the beginning of Tesla’s self driving project because he didn’t want to spend the money on lidar for all Tesla cars.
His “first principles” logic is that humans don’t use lidar therefore self driving should be able to be accomplished without (expensive) enhanced vision tools. While this statement has some modicum of truth, it’s obviously going to trade off safely in situations where vision is compromised. Think fog or sunlight shining in your cameras / eyes or a person running across the street at night wearing all black. There are obvious scenarios where lidar is a massive safety advantage, but Elon made a decision for $$ to not have that. This sounds like a direct and obvious outcome of that edict.
This kind of idiocy is why people tried to build airplanes with flapping wings. Way too many people thought that the best way to create a plane was to just copy what nature did with birds. Nature showed it was possible, so just copy nature.
To be fair, we achieved flight by copying nature. Once we realized the important part was the shape of a wing more than the flapping.
My vacuum robot uses lidar. How expensive could it be??
You need slightly more advanced lidar for cars because you need to be able to see further ahead then 10 ft, and you need to be able to see in adverse weather conditions (rain, fog, snow), that I assume you don’t experience indoors. That said, it really isn’t as expensive as he is making it out to be.
Regarding point number 2, I have no doubt we’ll be able to develop systems that process visual/video data as well as or better than people. I just know we aren’t there yet, and Tesla certainly isn’t.
I like to come at the argument from the other direction though; humans drive with eyesight because that’s all we have. If I could be equipped with sonar or radar or lidar, of fucking course I’d use it, wouldn’t you?