At Mashable Lance Ulanoff comes closer than most commenters to expressing my views of autonomous vehicles:
It’s time for some real talk about self-driving cars: they’re not coming around any time soon.
You won’t find a bigger fan of the technology than me. I love robots, autonomy and artificial intelligence. I can still remember visiting Carnegie Mellon University in Pittsburgh, and standing a few feet away from the car that nearly won the DARPA Grand Challenge in 2004.
But I’m also a realist — and despite recent promises by Uber and Ford, I know that self-driving cars are decades away from becoming a significant part of our lives.
Despite all of the hype we’ve been hearing lately (Ford producing fully autonomous vehicles by 2021, Uber being on the cusp of deploying a fleet of autonomous vehicles, etc.).
Consider the following.
- To the best of my knowledge there’s not a single street legal consumer level fully autonomous vehicle on the road today.
- Even if Uber has its fleet, it can’t use them presently in Pennsylvania (where they’re claiming to be putting their pilot program).
- Even if Ford starts producing fully autonomous vehicles, they would only be legal in a handful of states, e.g. Florida.
- It takes twenty years for the American passenger vehicle fleet to turn over completely.
- Software problems are different from mechanical problems. Every software problem is a failure of workmanship.
- We haven’t even touched on the security problems.
- Big, rich companies with deep pockets will be irresistible targets for liability suits.
- The first liability suit could deep six the move to autonomous vehicles for the foreseeable future.
What I hear you saying is buy Ford stock now, and sell before either the first, second or third time Ford again announces they are still five years away.
Glittering Eye, come for the sports talk. Stay for the stock tips!!!
I haven’t followed developments too closely because I think we’re talking about a technology that is a ways off, and the non-technological human elements don’t seem to get a lot of attention.
The notion that a primary user of the technology initially would be Uber probably solves the liability problem. Uber is going to be liable.
If there’s a chance that a billionaire can put a few million Americans out of work to be one dollar richer with this technology, it will happen. They’ll get on the phone to Congress and get extremely targeted tort reform done in a flash.
I agree it’s in the far future. If anything, this technology should be used for long-haul trucking first. Keep the people in the cabs to navigate the off-ramps, fueling stations and origin and destination travel, but the drudgery of interstate driving could start to be turned over to computers within 15 years (my WAG).
Google “Self-driving cars and the Trolly Problem”.
Trolley Problem
When I lived on a streetcar line in New Orleans, all of the streetcar accidents seemed to have been due to the fault of someone other than the streetcar driver, or due to some sudden occurrence like a dog running loose or a tree branch falling. The operating rules for a streetcar are pretty easy, but what to do about cars, bikes, pedestrians, dogs, and everything else?
The question I’ve wondered about is why haven’t we adopted autonomous railway engines? They’re an obvious candidate and highly realistic computer simulators have existed for generations. We don’t even have fully autonomous els.
The Trolley Problem is only incidentally about trollies. It’s a statement of a moral dilemma that makers of driverless cars will have to face.
Apparently, there is no trolley problem for the 16 year old with the brand new license. So, did I miss where the trolley problem worry warts are pushing to raise the driving age for adults only? That would be 26 by my reckoning.
Let me start the music for the song and dance.
There will be liability issues regardless of which decision is made in the Trolley Problem. BTW, there’s a robust debate on the prudence of raising the driving age.
@sam, I read the piece and thought it was an extreme example of a lot of risk assessments to be made by the driverless vehicle. What precautions should the vehicle take for what might appear to be a ball bouncing ten feet away, fifteen feet away, etc. And at what times should the ball be assumed to be a child? By the time, the car is asked to decide whether to hit the baby carriage or plow into the sidewalk, there will have been a number of decisions made with their own risk-trade-offs.
I imagine an algorithm in which safety can bet set any any point from zero to ten. At zero, the car indiscriminately drives to the destination, caring only for the comfort and convenience of the passenger. At ten, the car only gets up to 5 mph for it jerks to a stop and then jerks into motion again.
What I’m now reading is an explanation that the car will not exceed 25 mph and/or have a dedicated lane, i.e. manipulate the environment to reduce uncontrolled risks, kind of similar to a streetcar.