Autonomous Cars

[Greg Locock]

In my opinion anything short of L4 ignores human psychology. I personally would not use a L2/L3 lane keeping car, as I know damn well I'd be fast asleep in 5 minutes on typical Australian interstate roads. Most long distance journeys in Oz consist of 20 minutes of hassle getting onto the interstate or freeway, 600 miles of mind numbing boredom, and an hour of faffing around at the destination (if it is Sydney). I do use cruise control, I'm the annoying guy towing a trailer in the inside lane travelling at a constant speed just under the speed limit.

[Just_a_fan]

You guys gotta make up your minds. You now say it's understood you still have to be paying attention and have your hands on the wheel. Other times you talk of answering e-mails and doing work. Which is it?

Oh come on, be fair. The ultimate aim is to have technology that's good enough to be doing emails without the need to e watching the road ready to hit the big red button. At the moment, you need to be paying attention. That's because at the moment we don't have AVs, we have cars with some "driver aids" that are working out the issues with the intention of getting to AVs.

You seem to want it to go from the Model T to full AV in one stride or it's all a failure and pointless, won't work etc.

[AJI]

You guys gotta make up your minds. You now say it's understood you still have to be paying attention and have your hands on the wheel. Other times you talk of answering e-mails and doing work. Which is it?

Oh come on, be fair. The ultimate aim is to have technology that's good enough to be doing emails without the need to e watching the road ready to hit the big red button. At the moment, you need to be paying attention. That's because at the moment we don't have AVs, we have cars with some "driver aids" that are working out the issues with the intention of getting to AVs.

You seem to want it to go from the Model T to full AV in one stride or it's all a failure and pointless, won't work etc.

Apparently, as for some strange reason no one will make any technological breakthroughs in the next 20 years, these are the only two options?

I've got to say strad, you did a great job getting siemens to fluff goodwood in your mustang.

[Andres125sx]

I believe we are talking to people who don't take the job of driving as seriously as we do.
Also we are talking to people who are catering to the lowest common denominator.

Obviously, we´re talking about safety, and safety is only as high as its lowest point, because...

...and now with every idiot thinking it's their right to drive and at that to drive however they want of course there are problems.

You´re describing the situation yourself, but sill negate how dangerours are humans at the wheel?

Look like you simply can´t accept the reality you´re describing yourself because that would force you to accept AVs might be a good idea

Remember the video that was posted of Nico Rosberg in his Tesla? Remember when it scared him when it swerved too near that rock wall for his comfort and he was forced to grab the wheel?

This is a perfect definition of demagogy. You know that is NOT an AV, but even so you use it to bash AVs

[Fulcrum]

My Orwellian take on the topic.

For now, the discussion is when will AVs be "good enough" to trust, and in what capacity.

After that, when will AVs be statistically significantly safer, more efficient, etc... than people, and in what circumstances.

Then, the discussion will move to curtailing the individual's legal right to drive a vehicle.

Lastly, some jurisdictions may ban non-autonomous driving altogether.

I suspect the right to drive a car could become the new battleground of civil liberties; just like the right to bear arms is today in the US. Hopefully far less emotive though.

In all honesty, I'm both excited, and slightly frightened, by the idea of being driven around by an AI. Those that grow up with AVs as the default might think the exact opposite (people can drive - how crazy).

[Just_a_fan]

I suspect the right to drive a car could become the new battleground of civil liberties; just like the right to bear arms is today in the US. Hopefully far less emotive though.

Luckily, the US Constitution was written before the arrival of the motor car.

In the rest of the civilised world, the need for gun control is recognised by all sides.

[strad]

It probably will work fine until one of the cars has a computer glitch and then the Chaos Theory will set in.
I was looking up something entirely different when Chaos Theory popped up in a scientific paper that explained why all these carefully thought out traffic plans didn't work as they were planed.
You won't want to hear it but it also explains why their global warming/climate change predictions will go astray.

[Andres125sx]

It probably will work fine until one of the cars has a computer glitch and then the Chaos Theory will set in.

When autopilot for airliners were introduced there was a lot of people claiming how dangerous it would be arguing exactly this

[henry]

Personally I think the autonomous vehicle developments will continue on 2 tracks in the near term, nextv10 years.

One will be “out of town”, focused on long journeys on major roads with an objective of reducing the driver load ultimately to zero. These could be thought of as L3+ or L4- . They will be available to buy for personal ownership. Their sensor and processing needs will focus mainly on interaction with vehicles and road topology. Tesla’s latest “navigate on Autopilot” is approaching this functionality. They will incorporate sensors and mechanisms to ensure driver involvement.Cadillac’s eye tracking for instance. In a few years time (10?) this type of vehicle will likely get to L4 for the part of the journey from joining a major road network to leaving it.

The other will be “in town”. These will be true level 4 vehicles. They will be owned and operated by fleet organisations, possibly even manufacturers. Their sensors and processing will be more varied and require more servicing, hence fleet only. They will operate as ride hail, public transport, short delivery etc. They will initially operate in simple environments, orderly road networks, limited mixed traffic, good weather. Over time they will move to more complex environments as and when the operators think they can make money. Waymo in Phoenix is an example about(?) to go live.

I don’t think that it is likely that anyone will try to make money from L5 go anywhere do anything vehicles. There will still be driver operated vehicles for this need. They will benefit from the developments in “autonomous” research and infrastructure with aids to ease the driving burden. C Continuing the proliferation of blind spot, auto brake, speed limit detect etc. into optimum speed advice from roadside furniture, side road vehicle detection from vehicle to vehicle communication etc etc.

[subcritical71]

It probably will work fine until one of the cars has a computer glitch and then the Chaos Theory will set in.

When autopilot for airliners were introduced there was a lot of people claiming how dangerous it would be arguing exactly this

It's an interesting discussion. Aviation autopilots are very robust, even in light single engine aircraft, and are certified for every make/model aircraft they are installed in. They are also triple redundant in many cases. Does a car need the same type of certification/redundancy or do we let each manufacturer engineer what they think is best? Who would be the certifying authority?

Lots of questions and I personally think this will need to play out. If manufacturers cannot get it right I could see a regulatory agency needing to be created, like the ICAO (International Civil Aviation Organization) today.

[Tommy Cookers]

we know that autopilots are not computerised flight control
CFC design still kills people hundreds at a time (just as pilots did without CFC)
isn't CFC right now pretending to be what you lot seem to categorise as future L5 ?
in altitude acquisition mode the CFC will stall and crash the plane unless the pilot supervises it and exits AA mode

and will the computerisers write new laws for road use ? (and repeal the existing laws ?)
while they are doing this they might well write new rules for road design
non-AV road users could exploit/'game' the AVs
as smartarse cyclists and pedestrians do today in going against their red light when I have a green
eg what will my AV do when a non-AV doing a crazy overtake drives at me on my side of the '2 lane blacktop' road ?

[strad]

Confused.. Just lost two nice posts on the subject.
Condensed version... I agree that getting people out of the equation would help,,right up to the time when there is a glitch then chaos will take over...
I worry that tight wad manufacturers with be loath ,as always, to spend an extra penny, including on redundancy.

[henry]

....

isn't CFC right now pretending to be what you lot seem to categorise as future L5 ?

.....

I don’t think so. Level 5 for aircraft would control everything in getting the plane from airport terminal to airport terminal.

I don’t know much about CFC but it seems more like level 3, lane following and the like.

[Big Tea]

Personally I think the autonomous vehicle developments will continue on 2 tracks in the near term, nextv10 years.

One will be “out of town”, focused on long journeys on major roads with an objective of reducing the driver load ultimately to zero. These could be thought of as L3+ or L4- . They will be available to buy for personal ownership. Their sensor and processing needs will focus mainly on interaction with vehicles and road topology. Tesla’s latest “navigate on Autopilot” is approaching this functionality. They will incorporate sensors and mechanisms to ensure driver involvement.Cadillac’s eye tracking for instance. In a few years time (10?) this type of vehicle will likely get to L4 for the part of the journey from joining a major road network to leaving it.

The other will be “in town”. These will be true level 4 vehicles. They will be owned and operated by fleet organisations, possibly even manufacturers. Their sensors and processing will be more varied and require more servicing, hence fleet only. They will operate as ride hail, public transport, short delivery etc. They will initially operate in simple environments, orderly road networks, limited mixed traffic, good weather. Over time they will move to more complex environments as and when the operators think they can make money. Waymo in Phoenix is an example about(?) to go live.

I don’t think that it is likely that anyone will try to make money from L5 go anywhere do anything vehicles. There will still be driver operated vehicles for this need. They will benefit from the developments in “autonomous” research and infrastructure with aids to ease the driving burden. C Continuing the proliferation of blind spot, auto brake, speed limit detect etc. into optimum speed advice from roadside furniture, side road vehicle detection from vehicle to vehicle communication etc etc.

As this, but I would not be surprised to see another level slipped in to the system using static references on some roads to interact and fine tune the onboard systems.
I think governments, national and local, will not be able to resist sticking their ore in. A temporary redirect while there is something happing in an area for one, making sure they know exactly who is going into another area another.

The ability for them to instantly stop all traffic from entering or leaving a street, or just plain stopping all traffic would be irresistible to them.

A comparison, don't know who said it, but they were in security services I think, speaking about mobile phones.

Can you imagine the uproar if we said everyone had to carry a tracking device? Well now they not only do, but pay for it themselves.

[Andres125sx]

It probably will work fine until one of the cars has a computer glitch and then the Chaos Theory will set in.

When autopilot for airliners were introduced there was a lot of people claiming how dangerous it would be arguing exactly this

It's an interesting discussion. Aviation autopilots are very robust, even in light single engine aircraft, and are certified for every make/model aircraft they are installed in. They are also triple redundant in many cases. Does a car need the same type of certification/redundancy or do we let each manufacturer engineer what they think is best? Who would be the certifying authority?

Lots of questions and I personally think this will need to play out. If manufacturers cannot get it right I could see a regulatory agency needing to be created, like the ICAO (International Civil Aviation Organization) today.

I think that´s out of debate, obviously there should be a certifying authority, but I think it will be local for each country, as cars are not planes wich fly daily internationally. Then it will be a chaos if you want to go with your car to a foreign country if their standards are not the same, but hey, that´s happening even today in some circumstances just for the driving license so no reason to think this will be any different, absurd burocracy

They will be redundant too, anything wich implies safety of persons is redundant, and Strad, they´re redundant not because manufacturers are willing to spend twice as a choice, but because they´re forced to do so if they want their product to be homologated