Addenda to the book AUTO2010

More articles about the future,
previously published in small circulation magazines.

Plus a dynamite upcoming story predicting the next 100 YEARS!
As I add more, this page will probably become an index. ......updated 22 Jan 1996
the first one .....

"Why don't they make .....
a universal remote for accessories?"

You've been there. Your plane arrives late at night and it's cold and raining and you have to rent a car in a hurry. You finally get the car, and it's one you've never driven before. You start the engine alright, because the lock has to be on the column. Then you start groping: first for the headlights, then for the seat adjustment so you can reach around better. Then you grope for the interior lights so you can do a visual search. By then the windshield is fogged and you need to find the defroster, and heat, of course. The windshield wiper ought to be on the left control stalk, but which part of it? You drive over to the lot exit attendant, and can't find the right power window button. Finally you pull out onto the expressway, and realize the outside rear view mirror needs adjustment -- and you haven't a clue. Listen to the radio? Forget it, it's not worth the search!

Why do we put up with this? It's like the hassle of programming your VCR, except that the distraction consequences are a little more severe. Back in 1964 the SAE was able to establish the standard PRNDL shifting sequence, as a safety issue. And in 1977, the SAE J1138 "Recommended Practice" suggested that certain control functions be on the right or left side. But since then, the idea of standardized control locations has been a dead issue. Even if human factors engineers could reach a consensus, interior designers jealously guard their right to be different -- their freedom to confuse the public. There can be vast differences even between different models from the same manufacturer. In one quick survey, three model lines from one company had: seat adjustment on the door arm rest, the right of the seat, and the console; had climate controls above the radio, below the radio, and right of the radio; and had wiper controls on the left stalk, right stalk, and the instrument panel. And that's just a scattering of groups, not to mention the infinite possible arrangement of functions within a group.

All right, let them have their cake, but I'm going to propose that we eat it too. Let them put their manual functions wherever they choose, and let us have a simplified single auxiliary control. In fact, we need both, since there will be times when the old confusing controls will be more appropriate. But if ever there was a place for voice-actuated controls, this is it. We shouldn't have to take our hands off the wheel, our eyes off the road, or our minds off the traffic. Granted, most functions demand either more discrete triggers, than say, "horn," or finer tuning than coarse verbal commands allow. So, I suggest an additional 4-way thumb toggle on the steering wheel -- at least. The verbal command merely assigns different functions to the toggle.

Probably all automotive accessory controls can be reduced to an intuitive on-off, more-less, up-down, left-right, or front-back. As a rapid-response safety feature, I might suggest that the toggle have a default function of horn (in the center), turn signals (L-R), and dimmer (U-D). Then, for everything else, re-assign the toggle by any of many synonymous commands. If we may be so bold as to also have variable control, I'd like to have a computer-like track ball under the other thumb. That way, I could have more precise selection of variables such as speaker balance, instrument lights, mirror position, and radio station, and not have to wait for a time-dependent scan. The "Nintendo and computer generation" would certainly find all this quite familiar. I'd also like to have some head-up visual feedbacks, such as briefly replacing the digital speedometer projection with figures such as the radio station, the set temperature, or the cruise speed setting.

I've personally sampled some off-the-shelf voice-command systems, such as Lexus' car phone and the COVOX, and the potential is obviously awesome -- if not fully realized yet. The Lexus phone, for example, is an impressive show-stopper to the most jaded accessoryophile -- once the magic is set up. Unfortunately, you can't turn it on and start issuing conversational commands just yet, like, "Hello phone. Please dial 555-1212." First you have to read the manual. Then you have to punch in each of the 40 allotted numbers, and verbally record a distinctive phrase to go with each one. But, from then on, all you have to do is press a button on the steering wheel. A voice then says, "name, please," and you say, "mom," or "call home," or "police," or "Price Stern Sloan Incorporated," and it's done.

The COVOX Voice Blaster is a sort of do-it-yourself software package that you can set up to control your home computer by voice. It is a lot of fun to experiment with, but on page 18 of the instruction manual they point out that it should "....recognize between 75% and 95% of your commands correctly." I set it up with 20 typical automotive accessory commands from the table, and ran 200 trials, and got even less accuracy. When you think of computers, you expect nothing less than 100.0 % accuracy, and the auto industry probably wouldn't accept anything less than 99.99 % -- especially if it were a safety-related command. There must be better voice pattern matching algorithms just waiting to be discovered.

If there are just a few dozen standard command phrases necessary, it shouldn't be hard to simply ask for a "best match." Maybe what we also need is a universally accepted introduction phrase that clues the system into the language and accent of the speaker, like, "The rain in Spain....." The vehicle could also provide situational cues to eliminate less-likely possibilities. For example, if the vehicle clock indicates it's mid-day, then the command is probably not related to "lights." Or, if the sound system is off, then the command is probably not "speakers." Or if there is no speedo signal, it's probably not "cruise control." The auto has to be a fairly good environment for identifying commands consistently: fixed head location, good accoustic enclosure, and fairly predictable background noise, especially with the windows up and using a very directional microphone in the steering wheel.

These are examples of "speaker dependent" voice recognition, or the "trained" system. They most likely will not recognize the same phrase from a different speaker. For a universal system which will solve the problem of unfamiliarity in different vehicles, pre-programmed speaker independent systems are needed. However these are considerably more difficult to accomplish, and more expensive, the upper limit being 30,000-word dictation systems for $5000 and up. On the other hand, in anything electronic, it's taken for granted that technological advances and cost reduction come faster than the industry can adopt them, or the product liability lawyers can clear them. Even now, cost is not a major concern. The entry-level COVOX system suggests recognition of over a thousand phrases for less than $200, and in the mega-unit volume typical of the automotive market, the price would be a small fraction of that.

Wade Allen, president of Systems Technology, a vehicle dynamics think-tank, and member of the SAE's IVHS Human Factors Comittee, has already done research on audio control and feedback in automobiles. The purpose was to reduce driver workload due to even greater control demands from IVHS systems. As Wade says, "Voice control has to happen. There is little publicity on current research, but voice control will be critical in minimizing future driver workload and distractions. In fact, NHTSA should probably sponsor some research on the best way of doing it, so there could be a commonality of commands, the way we came to agreement on standardized symbols for manual controls. But maybe the greatest issue today is reliability, or consistent recognition of terms, independent of the speaker."

Right now, it's pretty much an OEM design problem, as it would be difficult to adapt an aftermarket device to different vehicles. At best, an add-on device might tap into the system by having an adaptor which plugged into the fuse block, where most of the control circuits pass through. But eventually, automotive systems will probably become more "open," or vulnerable to modifications. Multiplexing, on a 3-wire circuit bus, might allow access to everything, by tapping in at any light socket or connector. Using the example of video/VCR universal remotes, the desired signals could be programmed into the steering wheel VOX remote. Current programming practice is to either input known manufacturer codes, or "train" the remote by a learning cycle of running each device and recording the subsequent signals.

Naturally there are many questions to be resolved, such as whether it's harder to learn or adapt to, or what if a code phrase slips out of an unrelated conversation? Say you say to a passenger, "Save me a seat." And when you hit the dimmer toggle, your seat moves forward. Well, maybe the software requires a break in conversation before identifying a phrase, or it requires a "push to talk" button. We don't have to work out all the details here. Besides, it will take a major survey and research to get a wide consensus.

It's not enough to get one manufacturer to adopt such a systen -- for their own sales advantage. That's not the point. But it's probably expecting too much to hope for world-wide de facto standard, logical or not. I can't do it, but maybe you can -- or give me some excuses why not. I certainly stand to profit nothing from it one way or another anyhow. All I ask is that you name it after me: the "lazybutt'n."

This could turn into an "interactive magazine," if I can get your feedback as to future topics and other sources. Leave your e-mail address at:
If there are just a few dozen standard command phrases necessary, it shouldn't be hard to simply A w I s H o P k E g G c U _ W [ O W Q S A O I K h,             I & w & s #( o +( k g( g j( c * _ W [ O W Q S A O I K h,             y y y y y E y  y  y  y & y A y G y \ y  y  y  y 4 y y y  y <  y  y . y  y " y D% y e' y ) y !) y ) y ) y * y * y * y  y  y 4 y y y  y <     j# ontrol." The auto has to be a fairly good environment for identifying commands consistently: fixed head l  Arial accoustic enclosure, and fairly predictable background noise, especially with the windows up and using a very dire