All cars obsolete by 2020 according to Toyota

<div class='quotetop'>QUOTE(clett\";p=\"98618)</div>

I don't know about the US, but here in the UK a lot of people have an electricity tariff which charges maybe only 1/4 or 1/5 as much for electricity at night - it's often used for storage heaters.

That would certainly provide an incentive for anyone with an electric car to fill the car up at that time of night - another 75% saving on the "fuel" bill.

 

<div class='quotetop'>QUOTE(Jonnycat26\";p=\"98701)</div>

They would have to sense deacceleration and, with millimeter radar, closure distance, to modulate braking in "group" driving. It should be apparent that no single car can have braking too inferior or too strong.

What that would most likely mean is that your hypothetical BMW with "superior" braking might have to hit the tree anyway, as a high speed rear-end crash is usually more destructive than a lower speed frontal crash.

Actually, that is very much what happens now with "chain reaction" crashes. One driver may be able to avoid a crash, but the drivers behind are too busy jabbering on their cell phone or picking their nose or whatever, so they end up rear ending the person who managed to "avoid" the crash.

 

<div class='quotetop'>QUOTE(Jonnycat26\";p=\"98892)</div>

You make it sound as if crashes will increase with a fully coupled automated road system.

It should be obvious that a lot of research into - among other things - vehicle stopping distances will have to be done. It should also be obvious that any fully coupled automated system will need very tight PID control over things like rate of acceleration or deacceleration.

As with any automated and fully coupled system, it is impossible to reduce error to the mythical "zero" and more than a cursory examination of statistics will prove it. However, the whole point of automating a system is to dramatically reduce the variances to achieve a low and fairly predictable error rate (Determined by Weibull, ANOVA, DFFITS, FRACAS, etc).

As an example, a tightly coupled aviation autoflight system used in modern fly-by-wire airliners. There have been instances - so rare they become textbook studies on why they happened and how to avoid it - where the autoflight system glitched.

The glitch resulted in an altitude bust or inappropriate engine power setting. Revision of software has reduced this error scenario to a very low probability now.

Other "failures" in aviation automation were the result of human error. For example, inadvertently setting a knob for "F/S' instead of "FPA" so the airliner descended at 3,300 feet per second instead of 3.3 degrees flight path angle, crashing into a mountain.

Or a mechanic improperly servicing the primary ground busbar, resulting in a floating ground that glitched most onboard systems during a flight from Amsterdam to Boston, causing the display screens to display red X's instead of navigation data. In this situation, conventional electrically powered instruments also would have glitched, except there would have been no obvious warning to the pilots.

Your scenario is interesting though. Look at all the "chain reaction" crashes and rear-end crashes we now have on highways due to driver inattention. They're too busy using the cellphone, playing with their genitals, picking their nose, stoned, s*** faced, whatever, to be paying attention to traffic in front of them.

Do you for one moment somehow believe that wetware in the above setting is *more* reliable than tightly coupled autodrive systems? I suggest you pay a visit to NHTSA and look up the stats for those sort of crashes.

I would happily trade highly variable and unreliable response times for consistent and fairly reliable response times. With automotive application they're not starting from scratch as there is already a wealth of data and practical experience gained from aviation.

The systems now routinely used in aviation can be translated to automotive use. Which, BTW, your taxes helped pay for through DARPA and other agencies.

 

Thinking about the legal issue of automated road transport again, should the issue come to court it could go something like this:

"Toyota, you have been accused of negligent software development because this accident happened while a vehicle was controlled by your software. Can you prove that your system is safer than the average human driver?"

"Yes."

"Do you have an expert witness statistician to testify that accidents occur less frequently with your system than with human drivers?"

"Yes."

"OK then, you're off the hook."

Assuming this data was available, surely they'd be covered. Or am I well wide of the mark on this one? :?

 

<div class='quotetop'>QUOTE(Jonnycat26\";p=\"98992)</div>

Decision making. lol, you have not seen some drivers here in El Paso. Somehow I do not think they were deciding much. hhhmm let just change lanes with no blinkers 1' infront of another driver. I especially like the guy who was traveling at 80 MPH on the shouder because traffic was going 20 mph due to the accident on the should of the road 1 mile ahead. I wish I was there while he was explaining why he was on the shoulder to the cops. My morning and afternoon commute is an education in idiocy.

 

<div class='quotetop'>QUOTE(clett\";p=\"98593)</div>

Nice ideas clett. First though, I would keep my BP stock. They are the world's largest producer of photovoltaic cells. So let's take your ideas a step farther. Instead of the liability the store would face with a customer using an actual plug, particularly in rain, snow, fog, etc., just use magnetic induction to charge a car. Pull up to a special pole (protected by a concrete bumper of course) and the red light on the pole changes to green when close enough (like in laser car washes). Now the flat roofs on any supermarket might be able to carry quite a charge for the store as well as customers. And of course the supermarket wants the customer to spend as much time (and money) in the store so if customers spend 30 minutes or more shopping, even a small charge at 5KW, should be enough to pay for the trip there as well as get a little extra charge. That's a primary reason why Toyota doesn't support plug-in charging, the liability of the actual plug.

 

<div class='quotetop'>QUOTE(Jonnycat26\";p=\"98992)</div>

I guess you haven't toured any large modern chemical plant, especially the control room?

Honeywell had that technology - admittedly at a very high cost - back in 1980 with their TDC 3000 running LCN and interfacing plant sensors and control loops through Hiway Gateway. The current iteration is Experion Process Knowledge System.

Fisher/Rosemount had similar systems too, along with Bailey (INFI 90 series) in the early to mid 80's. The current version of the Fisher/Emerson solution is DeltaV PlantWeb.

I started off in the "good 'ole days" of discrete control using 4-10 mA loops, primarily PID and Cascade loop. Now everything is an "open" bus like Industrial Ethernet, Foundation FieldBus, ControlNet, or DeviceNet.

Deterministic control and "fuzzy" control loops have been widely available since the early 90's. Plants that have upgraded their control strategies to the 21st century have experienced not only increased uptime but much improved process safety, especially in chlorination cells.

A nagging problem still remains in that the process Safety Monitor will do a preventative shutdown when it senses an out-of-range value or even a bad bearing. The operator will override the safety shutdown and then everybody is surprised when there is an explosion.

Similar safety and control measures are applied to fly-by-wire airliners, such as the Airbus series. For example, the aircraft is protected from excessive pitch input by Alpha Floor. The pilot still has the option to disable the safety systems and manually fly the airplane into the ground.

That's why that Airbus A-320 crashed at the airshow in France around 16 years ago. The FMGS wouldn't let the pilot execute a low-and-slow flyby, so he disabled all the safety systems and flew manually. He stalled the aircraft and crashed.

I personally like the idea of modern airliners and naggy vocal reminders when the pilot forgets something : "Flaps ... terrain ... sink rate ... glideslope ... PULL UP!!"

If you still honestly believe, despite looking at all the stats from NHTSA, IIHS, and others, that the wetware behind the wheel is safer than an automated system ... you have my sympathy. I also guess you don't get out on the road much.

 

<div class='quotetop'>QUOTE(Jonnycat26\";p=\"98992)</div>

Ah, I am a great follower of Moore's law!

Take a look at what Sony's Qrio and Honda's Asimo are already capable of.

They can recognise people, faces, voices, instructions and objects and move around safely among people and within their environment - and that's with complicated bi-pedal motion, 4 wheeled driving would be trivial by comparison (at least in terms of control of motion).

But of course, as you say, the problem would be decision making. Should the car pull out past an obstructing stopped vehicle, or would it be safer to wait for it to move? Should it break the speed limit to make a safer overtaking manouvre? Does that look like an accident up ahead? I suspect that in reality only licensed drivers will be allowed to use such systems, and that in situations such as these a buzzer would sound and the driver would then have to wake up and take control for that section of the journey - rather like an extended version of cruise control. Version 1.1 would likely be a motorway only cruise control - switching off automatically by GPS at the sliproad.

I don't believe the problem will be a lack of computing power when it comes down to it, but the software that runs it. (let's face it, a spider's brain copes with much more difficult visuo-spatial tasks than cutting through rush hour traffic! It's just efficiently programmed.) You can do a lot with a teraflop, but I'm the first to agree that the software challenges for any of this are non-trivial to say the least!