Comment by paganel
3 years ago
Not all driving outside city areas happens on the highways.
My brother lives in the countryside (somewhere in the EU), and, as a driver, he shares the paved road just outside his house with the village’s cows (including his two cows). I don’t see any non-AGI system being able to negotiate that, as at times is difficult even for me, a reasonably AGI system, to make sense of it all when I encounter a herd of loose cows on the road.
Sparsely populated rural areas are really the only place where personal autos make sense. Long distance highway travel is better serviced by train. And transportation inside the city is better serviced by literally anything except a car (bikes, e-bikes, subway, trams, scooters, walking).
Even cities with decent public transport networks (such as Prague) tend to have a problem with tangential relations. Most lines go from some periphery to the center and to another periphery. If you need to travel between two peripheries, you have to be somewhat lucky to have a line connecting them at your disposal. Going through the center and back to the other periphery is possible, but usually takes too long. (e.g. 1h 30 min instead of 15 minutes by car).
> Sparsely populated rural areas are really the only place where personal autos make sense
This might be true, if rural areas and densely populated uran areas were the only 2 states, but there's a lot of intermediate density in between
Having kids makes car sharing very unattractive. Finding car sharing with enough (and high quality) car seats can be quite challenging. And as much as I like public transport, with kids using the car is often much faster.
How do you get around the sparsely-populated rural areas at either end of the highway?
https://www.youtube.com/shorts/fRAAf83QzR0
And this is just a single clip. The huge advantage Tesla has is their massive amount of training data. If it happens in the real world, they probably have a clip of it and can train on it.
Sheep are worse. They run from one side of the road across you to get away from you.
Add to that the vagueness of where the road ends and field/bank begins for many country roads.
You don’t need AGI to do edge detection it’s purely a vision problem
This is also why Tesla switch to using 'drivable area' rather than strict road boundaries.
Oh come on, it’s just collision avoidance, you don’t need general AI for that
Collision avoidance that needs object permanence including theory of mind (for cows). How hard could it be?
What? You don't need to model a cow's mind to steer a car around it.
This discussion is so absurd. :)
1 reply →