To be fair, if they crack automated driving it'll revolutionize public transport too. Imagine buses routed dynamically, and sized perfectly to match demand. Or other automated, private cars on the road yielding to public transit vehicles.
So basically something like personal rapid transit, first deployed ~50 years ago? And for which virtually every implementation has disappointed?
> Or other automated, private cars on the road yielding to public transit vehicles.
It's called "transit signal priority", and has been around for at least 20 years.
Signal priority is only at signals. I'm talking about private cars changing lanes to let public transit pass through. Human drivers will only do it for emergency vehicles, but you can make it a condition of licensing for a private driverless vehicle.
The first PRT system was (and still is!) completely automated.
We don't need self-driving to do this, and in fact many localities have already tried this with human drivers, and it's been an abject failure. Dynamic-dispatch public transit has been tried!
The results have been low usage (because it turns out the predictability of a timetable and consistent route is a feature, and not a bug!), with high wait times and long trip times.
Heck, the major ride hailing companies have done pilots on this front (ride hails as last-mile supplements to fixed-route public transport) and it's been a pretty thorough failure there as well.
In urban planning circles there's the term "gadgetbahn" to derisively describe high-tech, flashy transit projects using unproven tech that ends up being far worse than normal proven technologies. Dynamic-routed public transit is like that, but with even less "gadget".
> and it's been an abject failure.
For the product being delivered (low speed, low latency, available only in specific areas), public transit is criminally expensive. I'm not surprised people prefer driving in suburban areas. If it was 1/10th price, or free, it becomes a lot more attractive. Free is impossible if you have to pay salaries to human drivers.
Driverless vehicles don't have to be 4-seater cars. They can be much smaller, like the Smart Car, or a single person enclosed three-wheeler.
This is true - the problem is that dynamic-dispatch transit is even more expensive on a per-passenger or per-ride basis, and both speed and latency are actually worse than status quo public transit.
It'd be one thing if the result of these projects have been "this is great but expensive", but the result overwhelmingly has been worse than fixed-route bus transit across every major metric.
Autonomous cars can improve the cost problem, but can't dramatically change the calculus on speed and latency. Likewise, autonomous cars can also solve the cost problem for buses, and better!
Even if you remove the human driver and the need for a 3-ton piece of metal powered by an ICE?
> both speed and latency are actually worse than status quo public transit
All the ride-hailing apps seem to appear to disprove this. They're more expensive than public transit for this very reason.
> Likewise, autonomous cars can also solve the cost problem for buses, and better!
Hear hear!
We've had self-driving buses since 1875, they're called trams.
Imagine seeing vehicles automatically moving to the side of the road to let an emergency vehicle through. Or the subtle nudges and lane shifts to ensure traffic doesn't reach a standstill.
It's a great use case for a hive mind ML/AI.