This Tesla Robotaxi demo video is a mess.
Watch as the car makes a left turn from the wrong lane, ignoring a red light. The safety operator steps in, and the car comes to a stop… right in the middle of the intersection.
Eventually, it completes the illegal turn after blocking traffic for 45 seconds, which raises the question, what exactly is the safety operator there for?
Any car with an automatic emergency stop feature is safer than an “average human driver”.
That’s a big problem with all these self-driving car statistics: Self driving cars are usually very new and outfitted with top-of-the-line features, while the “human driver” they are compared to drives a much older and cheaper car, often without many of the security features that new cars are required to have and often not even maintained properly. Doesn’t really say much about the driving capabilities of a self-driving car.
Sadly, actually comparable statistics are impossible to find, since accident statistics aren’t usually collected on a per-car-type-and-age basis.
I’ve been thinking about exactly this… you could probably get an apples to apples comparison, by comparing only a specific model year.
IE, how did every 2023 car perform in the year 2024? Then, parse that by the vehicle’s price, or maybe by its propulsion type, maybe both. THAT would be how an honest study of the question would go.
Of course, Tesla knows that, they can afford really great data scientists. Makes you wonder why they run their obviously flawed safety study instead.
Tesla even has the best data source for a control group in house: those of their customers who don’t use autopilot.
They use identical cars with identical security features and Tesla has all the data. It would be trivial for them to compare those numbers.
As you say: there has to be a reason why they don’t go for this extremely low hanging fruit.