Published by ABC News on 24-Jun-2016
When driverless cars hit our streets, they will rely on a combination of sensors and sophisticated programming to navigate safely.
A study published today in the journal Science found we want the cars to act in different ways depending on whether we're inside them or not.
The participants would apparently not rush out to buy a car that sacrificed their own family's safety to save pedestrians [or cyclists].
• Scientists asked people what driverless cars should do in a range of fatal accident scenarios
• They made different choices if they were inside the car, or observing
• The findings illustrate a thorny issue for programmers: save lives overall, or prioritise passengers?
. . .
I'll put my money on some one being hit by a driverless car, while using a pedestrian crossing.
What about wireless blackspots, e.g. Adelaide Hills or the Great Dividing Range? Will cyclists be at risk from unreliable vehicles? Or as drivers forced to revert to self-drive, will the now inexperienced drivers create hazards?
Driverless cars 'could be on roads by 2020', Volvo predicts ahead of first Australian trial
Published by ABC News on 7-Nov-2015
. . . Mr Shearer said more than 240 legislative changes would have to be made before the vehicles could be made available for the public.
He said the vehicles could also become unreliable if travelling through wireless black spots. . . .
We have many wireless black spots in all of Tasmania, which is very hilly.
> He said the vehicles could also become unreliable if travelling through wireless black spots
So how does that work.. GPS degrades so a less fine grained approximation as to "where am I?" kicks in which is mobile phone tower triangulation. Both of those are less reliable than the cameras mounted on the vehicle itself and what other proximity sensors that are employed.
Wifi out? I don't think we are crashing into the breaking car in front .. nor drifting lanes ..
My understanding is that they primarily rely on their camera, not their GPS. So they could still work without GPS.
That said, I'm sure any initial roll-out will be restricted to city areas with good reception.
The thing that bothers me thee most is the programmers need to make an ethical decision on in a particular situation is it better to kill the occupant or the pedestrian/cyclist. How does a car company make that programming decision? What would be the legal ramifications if it was determined the car company had programmed the cars to kill pedestrian/cyclists over the occupants? The crash no longer becomes accidental as there is no human intervention at the time and it is down to the programming of the vehicle. Minefield.
Perhaps we read the same article today. If morals and ethics can be made logical then the machine can execute said logic. I guess the coders need some decision making nirvana.
Even if easier choice of save the cyclist's life and perhaps injure a passenger (protection with metal body, seat belt and air bags), expect car-centric public would clamour for their maximum protection. Also expect vehicle manufacturers would support this (an example of bull-bars).
In some ways, I don't think it could be worse than cars with drivers. Plus there is opportunity to put in all sorts of mandated behaviour by legislation.
Also, I think situations which are a choice between two human lives are extremely rare. It's hard to imagine one where the car is not initially driving recklessly. e.g. when driving my car I have hit a kangaroo, and I made a conscious decision to not swerve and not endanger my family, even if it meant killing the roo (even though it survived). (My rule when driving, which I have drummed into my kids, is to *never* swerve for an animal. Brake but never swerve.) But I was driving 100 and the roo jumped on to the road in my path. In areas where a child might run out, I shouldn't be driving 100 in the first place - and no driverless car would.
Does anyone remember the film "Westworld" (1973)? [Wikipedia link]
Driverless cars could be hacked.
Manufacturers have moved towards air-gaps to avoid hacking sometime ago.
If anyone else here has done Electronics they will know things can go bad and parts don't last forever, think of it as big skaletrics cars haha, lasers , sensors, constant calibration , I wouldn't never trust a machine with my life in that way, that's why we have safety fences on robot work stations. Another 50 years maybe, walk before you can run imo. Yes they can definitely be hacked as with any wireless tech.
If they must do it, Do it the Liberal way with Copper and Steam haha, cant even get the internet right let alone driverless cars