No, you've got it backwards. The point of all of the above comments is that driving in cities and densely populated areas is a much more difficult challenge than driving down a straight open highway in the middle of nowhere. That's the easy part, the hard part comes when there are humans and all sorts of obstructions and road/land features to deal with.
Freeway driving is pretty straight and forward. If you can't keep from running into a stationary object, you probably can't drive in the city either.
If an object appears on the road, the autonomous vehicle can easily calculate whether it's safe to steer left, or right, or hit the breaks. A human will make a roughly random choice.
Autonomous cars don't have an advantage because they don't exist. They may have an advantage one day when they are finally ready (but that may always be 5 years away)
> If an object appears on the road, the autonomous vehicle can easily calculate whether it's safe to steer left, or right, or hit the breaks. A human will make a roughly random choice.
Even with a 360 degree view, it still has to recognize the object, classify it correctly and make a decision. All tasks which are difficult for a computer to perform reliably enough to be allowed on a public road.
The Uber car had a 360 degree view but it still ran over Elaine Hertzberg, and as far as we know didn't even stop after it hit her. The Tesla Autopilot has 360 degree view but still tends to crash into barriers, tractor trailers, street sweeping trucks and fire trucks.
Fast driving necessitates predictability. Unrestricted stretches of the autobahn are not killzones because the speed breeds discipline: lanes are changed with indicators on more often than without (barely, but still..), passing lane rules are followed almost religiously, compared to what you see in America (even though drivers constantly complain about imperfections) and so on. The fastest roads of any jurisdiction would its easiest for a robot car or truck. Slow environments on the other hand are full of ad-hoc improvisation and weird traffic participants like the occasional flock of sheep, horse team or a biathlon squad on rollers.
Honestly I would think that automating a slow moving vehicle in the city would be a much harder engineering challenge. City driving can be a lot messier than driving on the freeway.
As the recent Tesla crash showed, even on a divided highway there are plenty of opportunities to cause death and mayhem to the autonomous car and the surrounding vehicles.
Garbage trucks are big and heavy, so if things go wrong they can go badly wrong, such as the incident in Glasgow that killed six people[1].
It has. The EasyMile EZ10 (a passenger shuttle) has already been deployed in a lot of places, for example.
This revolution should have started with mundane tasks, such as city street sweepers and perhaps garbage trucks. Vehicles that move slow on a predefined path in the night. Then advance to faster moving objects like taxis.
Just like the DARPA Grand Challenge that started this mess. Many of those vehicles were industrial usage, not consumer.