murray928

How much fun would that be ? Who wants these driverless cars besides Google ?

Speedtoys

They will still require drivers...just the press and the 'average joe' likes to think they will be driverless.


As much as I at times enjoy/leverage autopilot flying, autopilot in a car can have it's advantages as well.

dr bob

Biggest risk in a driverless car environment will be non-driverless cars. Jeff's reference to autopilot in the plane is interesting. In a way-back-when project, I spent time working on some of the software used for the DC-10 simulator at Flight Safety, under contract to M-D (now a part of Boeing).

The idea of full fly-by-wire was ugly to those used to having three professionals on the flight deck. Even though we could readily demonstrate that the redundant auto systems were more reliable than cables and pulleys, it was a seriously hard sell. Today, commercial airliners can literally fly themselves, but still depend on some ground-based systems to work correctly. Sometimes it's really stupid stuff that requires pilot intervention. So long as there's any risk of something needing that intervention, we'll have pilots driving the planes even when 99+% of the time the pilots are not needed.

With the biggest risk to driverless cars being the non-driverless cars, we won't be seeing actual driveless operation until manual driving is eliminated on shared roads.


In my spare time, I write software that automates power plant operation. It's a cakewalk compared with driverless cars, mostly because I can easily wrestle critical control from the operator if they try something stupid. I'm not sure that will happen in cars until brakes and steering are wire-drive without mechanical connection from the driver. Biggest danger in power plant automation is lack of instrument and valve maintenance.

Red Flash

Although I have no technical background knowledge, I have this common-sense feeling that what Dr. Bob says make sense.

Given the situation, I foresee a time coming when manually steering cars will need special permits for operation and will not even be allowed on interstate highways.

I wonder what will do the market for exotic cars? Maybe for really nice specimens not much, but for many, the market will just get to small and value might drop?

Speedtoys

Yup..automated systems often find disagreement between both paired systems in a cockpit, and hand the plane to the pilot.

That's when many of the crashes happen, the pilots are so rusty in procedures, THEY end up crashing the plane.

That will, IMHO, soon be a day with cars. As bad as we are today at driving, when automation hands the car over to you, you'll be to unprepared and experienced to do the right things in time.

Not to mention insurance liability..no..you'll need a human there to be the pilot in command, period.

Speedtoys

Nah...just like automation-less planes can still fly the airways.

There is a built in _requirement_ to be able to operate without automation...cars will follow the same route.

GT6ixer

Big difference between planes and cars is in the spacing between other vehicles and obstacles. If, for example, an airplane experiences an autopilot hardover (i.e. uncommanded control input), the pilot can manually recover the airplane albeit with some loss of altitude or lateral course. If the same thing happens in a driverless car, chances are by the time the driver reacts, the car wiull have run into a another car, a barrier, a ditch ect. IMO the algorithms and redundancy that go into the driverless car have to be on par or better than commercial aircraft autopilot systems. I work with FAA certification everyday for my job, and if the autonomous cars gets regulated at the same scrutiny of aircraft the Feds are going to have to subsidized the auto manufactures. Otherwise it will be prohibitively expensive to certify. Likely much more than crash testing.

Speedtoys

Thats -a- scenario.

The other being trimmed straight and level flight...or on open road.

I dont think auto will go the way of Part23 costs and hassle, look at the technology of auto-braking...effective, and the packages are getting pretty cheap to install per car compared to 5yrs ago.

But..automation based accidents in cars, will create a huge problem for insurance companies, because it will be so much easier to sue the manufacturer all the time, where today, auto systems are rarely put to question in an accident.

dr bob

Same concerns with the power plant drivers. Old-school drivers say poo-poo to the automation stuff, that the newbs need to learn things the way they did. Like everything else though, the generation-me (old farts...) crews don't relate well with new candidates who are used to having the computer do everything for them. Plus, folks who grow up with automation support don't have a clue what to do when things don't go quite as planned. Kinda like 60's stick-shift cars with manual drum brakes, no power steering and no AC. Are auto transmissions better? For most folks who've learned to drive in the last 20 years, they are a functional requirement.

For the plant and pilot automation, mandatory simulator time is part of the answer. Plus we examine every little oooops, and add the conditions and required responses to the "what if?" list. For a client a couple years ago, transitioning out of coal into gas turbines, their simulator expert had a training protocol that included inducing specific failures to test for operator response. I asked him for the list and the correct responses, so I could add them to the control package. Then asked him to think up some more. At some point that will be the protocol for refining driverless cars too. Keep learning from our mistakes as well as our successes, continually refining the strategy towards perfection. Keeping in mind that the approach is asymptotic, always approaching but never actually achieving perfection. One of the easy mistakes that will be eliminated early is drivers in non-driverless cars.

Speedtoys

Pilots retrain every 6mo...the issue was not USING their hand flying skills, not a lack of training.

dr bob

But there are limits to options. Nobody wants to risk their passengers, aircraft, plant equipment, reactors, etc., to allow the pilots/drivers/operators to get real-world operating/flying/driving experience with any of the safety systems disabled. (see: TMI, Chernobyl et al...)

Speedtoys

Manual flight without autopilot and auto thrust on...

..has no guardrails...you will get alarms if you try to stall it, but you can...if you try to roll it over, but you still can, or dive it into the ground..and you still can.

When you turn off George, you can do whatever the hell you want.

PaulD_944S2

"Big difference between planes and cars is in the spacing between other vehicles and obstacles."

One thing to consider is that if/when the vehicle's auto pilot fails and the human driver takes over, the immediately located vehicles will respond with actions of moving away or rapidly reducing speed.

Part of the driverless car plan is to have vehicles electronically linked together and aware of each other's kinetic parameters to control exactly what you think might happen when auto pilot shuts off.

dr bob

Deciding whether to let a [pick your driver] go on with no belt, no suspenders, no harness, no lanyard, etc. is a decision I leave to insurance underwriters. When someone proposes a real-world exercise with all the safety-net features disabled, I gently suggest that they clear it with their insurance group, in writing. For self-insured utilities, a written authorization from senior management is adequate. Hasn't happened yet.

Driverless cars will be the same. Like Flo's little plug-in tattletale dongle, you'll have a separate premium surcharge added to your bill each time you decide to take over from the system. Or write a letter for each time you do it, explaining the need. Costs will kill operator/pilots driving unless a real need is established. Real-time data recording is here already, so less chance of fooling the system.

Get used to it, it's coming. Just a question of when.

XS29L9B

For a multitude of reasons, eff-NO!

Can't think of any real benefit of having a "driverless", or automated vehicle. None.


George Orwell, here we come...

And to the OP and their idea of a driverless 928, surely the costs associated with adapting an "old car" to "modern" tech, would be astronomically prohibitive. But what a controversial topic, anyways.