How will self driving cars affect motoring

Best will be driving out to the pub, getting slaughtered and then back to the car and asking the car to bring you home after which the car will get thick with you like a version of KITT.

and if there was an accident between a manual car and one of these, will the judge always side with the computer?

I saw this video yesterday and thought . "wow .. thats great"

Then I thought what if the blind guy is out in the middle of nowhere and the car says "ERROR 2AB5" he's f*cked.

Or if his car drives through some sh*tty neighbourhood and kids start throwing rocks at it.

In a perfect world it would be great, but we don't live in a perfect world.

I think automated systems could be great for motorways and would greatly increase capacity. Motorways are boring as it is, something like this would save fuel, make the mundane commute more bearable and a lot more safe.

I don't think this will happen any time soon. Human brain is still superior to machines in terms of being able to act instinctively and it can solve dangerous situations better and faster as it can predict unpredictability. While I expect certain automation in the area of public transport, I believe cars will still be driven, for fun if anything else. Look at car racing, while pretty much automated, it's still humans making the last decision.

what happens if a horse jumps in-front of the car?

markpb wrote: »

I think people are over-estimating how long this will take. Companies have been building cars that can successfully drive navigate highway and urban traffic for years - the only remaining major problems are societal and legal.

Look at the advances in technology in cars over the last ten years. All it needs is for this to continue, gradually phasing out little things that we used to be responsible for until one day, there's nothing left. Right now, a high-end car can maintain a fixed speed, maintain a safe distance from the car in front, brake if the car in the front stops suddenly, maintain it's own lane, read speed limit signs, be aware of pedestrians standing on the side of the road and maneuver round obstacles that appear suddenly. Research cars can merge into moving traffic and negotiate four way junctions.

+1

Technology for self driving cars has been around for years. Finding who'll take the blame when it goes wrong is what's holding self drivings vehicles from the roads

Talk about taking all the fun out of driving :eek:

Chris_Heilong wrote: »

and if there was an accident between a manual car and one of these, will the judge always side with the computer?

The data will all be logged and the exact circumstances, including the position of both vehicles can be precisely replayed over and over as required.
I would assume that most such car would come with multiple cameras by default anyway.

haminka wrote: »

I don't think this will happen any time soon. Human brain is still superior to machines in terms of being able to act instinctively and it can solve dangerous situations better and faster as it can predict unpredictability.

In reality a computer in any such scenario will pick the better option 99% of the time. To a certain extent, instinct is part of the problem with human problem solving and can cause us to pick a solution that's not optimal based on our personal biases.

A computer can crunch the scenarios in a fraction of a second and pick the best one. Undoubtedly this will result in Daily Mail headlines where a computer decided to crash into another vehicle containing a blind, deaf and dumb paraplegic rather than drive itself off a cliff, but such is life.

no1beemerfan wrote: »

Thanks be ta fcuk I'll be dead and gone before robots take more and more control away from humans and we as a race end up dumb and lazy slobs!

You think we're not already?

Chris_Heilong wrote: »

what happens if a horse jumps in-front of the car?

The car will pick up the horse before any of the occupants even see it and will account for the scenario of the horse appearing on the road. There is no such thing as something that appears on the road "out of nowhere".

Driving is for the most part actually a fairly deterministic and uncomplicated thing. You're going from point A to point B, following a set of rules. The complication appears because there are people involved. But even then it's still fairly deterministic because there are only a set number of things another vehicle on the road is likely to do, so you can account for them.

Computers can't keep track of 30 other vehicles? Sure they can. They can keep track of hundreds, all at once.

The main barrier to their introduction is that we don't like ceding control. We could build self-flying aircraft at the moment which are far safer than human-flown aircraft, but nobody will get on a plane if there isn't a human at the helm.

The Google cars on their initial test drives clocked up over 100,000 miles and the only intervention required was a single emergency brake. How many of us can claim to have done 100,000 miles without making a mistake? None of us.

But even if automatically-controlled cars dropped road deaths to (say) 10 per year, the argument will always be that, "Those people wouldn't have died if a human was driving".

In reality the innovations developed in self-drive cars will be brought in gradually to augment the human experience. We're already seeing these - cars which park themselves, cars which do an emergency brake by themselves about 100 times faster than the driver can. SatNav will become collaborative, speaking to eachother over the cell networks such that your satnav will alert you in real-time to what's going on - showing you that there's an ambulance approaching 500m from the left, or that the guy in front intends on turning left in half a kilometre so you should change lane.

These changes will creep in so gradually that people will applaud their improvement of the driving experience until cars are by default fitted with the technology to drive themselves. Then it's a matter of just hitting a button. Then self-driving cars become so normalised that most people will prefer it for most trips, and taking the wheel yourself will be considered kind of dangerous and frowned upon.

From there it's only a matter of time before you can drive a car without insurance or a licence so long as it's on auto-mode, and people will stop getting licenced or insured at all and using the auto mode for all of their trips.

seamus wrote: »

SatNav will become collaborative, speaking to eachother over the cell networks such that your satnav will alert you in real-time to what's going on - showing you that there's an ambulance approaching 500m from the left, or that the guy in front intends on turning left in half a kilometre so you should change lane.

I think this kind of networking will be key to whether automatic driving really works. A few years ago I read about research into ad-hoc car networking, where all the cars within a few hundred metres of each other would form a local network and inform each other of their position, speed, immediate destination (e.g., turning off at the next exit) etc. The idea is that the vehicles would automatically negotiate to find the most efficient way for all to travel. It also reduces the need for each car to blindly monitor others as the intentions of each are known.

Turbulent Bill wrote: »

I think this kind of networking will be key to whether automatic driving really works. A few years ago I read about research into ad-hoc car networking, where all the cars within a few hundred metres of each other would form a local network and inform each other of their position, speed, immediate destination (e.g., turning off at the next exit) etc. The idea is that the vehicles would automatically negotiate to find the most efficient way for all to travel. It also reduces the need for each car to blindly monitor others as the intentions of each are known.

There's already a satnav app which kind of does this:
http://www.skobbler.co.uk/
The app reports via 3G how long the car has been sitting in traffic and then nearby users can use this information to automatically re-route.

The problem I think is that it's probably ten years ahead of its time and requires far too much manual intervention to report problems. Unless you have a critical mass of people using these collaborative systems, it's not that useful.
In order for it to become popular, the system will have to be based on open standards (so it's not manufacturer-specific) and standard SatNavs will need to have general access to cell networks to make the updates.
So it's a good bit away yet, but definitely on the cards.