Slashdot Mirror
Cringely, Cars, and Networks
Boiled Frog writes: "Cringely's latest article talks about Telematics, the art of putting computers in cars. However, the more interesting part is near the end where he talks about mesh networks where every car would have a router in it. I could see this extending digital cell service and mobile network connectivity far into rural areas."
I, personally, am all in favor of Transit Area Networks, which would be ad hoc wireless networks set up between cars. They could communicate GPS information, and if the driver wanted to, voice data. In case of a crash, the cars would each have a record of what all of the other cars in the area, etc. did.
There are horrible security problems that would need to be worked out, but I've often wanted to be able to beam an "I'm sorry" onto the HUD of the driver I just cut off, or something like that.
There should be a moratorium on the use of the apostrophe.
Max V.
NeXTMail/MIME Mail welcome
This would be heaven for all those script kiddies out there willing to expand their destruction to moving objects as well!
Imagine your dashboard flashing "I OWN3d J00!!!" moments before the gas and break pedals switch functions.
Looking for people to chat about multicopters, coding, music. skype: gtsiros
I actually study Telematics, and it's not the art of putting computers in cars. It's simply the application of computing to (long) range communication.
The IETF has already thought of scenarios like the one described (a router in every car). Look at the Mobile Ad-hoc NETwork working group.
I guess this gives a new meaning to the term "network traffic."
The next comment I write will be ready soon, but subscribers can beat the rush and see it early!
Cringely builds his argument for a widespread, car-based wireless network on the premise that the storage required by cars frequently disconnected from a network is an insurmountable problem, given the inability of Hard Disks Drives to survive in the hostile environment of a car. He believes that this problem will not be resolved by HDDs designed to better cope with that environment because the HDD companies can only afford to invest in research that will pay off within a year whereas the car companies plan four years ahead.
IMHO, it's a bit short-sighted to focus exclusively on HDDs; Flash memory makers are currently making great strides in producing chips that, in capacity, compete with miniture HDDs. Their primary financial motivation for this is the perceived huge market for personal MP3 players. I read one article a few months back that predicted a real head-to-head battle between Flash memory and IBM's tiny HDDs.
If we're going to be seeing Flash memory with several GBs capacity, I don't see why they shouldn't be used within cars.
Also, I don't see why the 4 year planning cycle for a new car should be such a problem; that time covers the design process for the car as a whole, no telematics system would be so intrusive as to require being part of that process from Day One. Indeed, it should be something that can be integrated within existing designs.
I'm wary of questioning Cringely's ideas because he does seem to have good sources on this but the direction he's taken that info doesn't seem to have been thought through properly.
Also, it's hard to accept his technical credibility when the software he uses for his site's forum is so damn tacky.
I work with Telematics as a software systems architect.
The main problems we encounter are:
Coverage.
Only AMPS can provide any kind of decent coverage in the states. This means 1200 BPS and heavy error correction. We use everything from DTMF tones to stripped down V.27 (fax) data bursts to get the message across. No. Sattelites are too expensive.
Driver attention.
If you are driving a vehicle, you should not play around with an interface. There are also regulations on what you can put on the dashboard when it comes to screens etc. Services has to be kept extremely simple. We are just learning to do voice recognition in an automtive environment with noise, radio, screaming kids etc.
Tech cycles.
A car is supported for approx. 10 years by its manufacturer. Which mobile standard should we use, given that GSM, GPRS, or CDMA gets the right coverage? Forget 802.11 where the vehicles act as nodes in a network. 802.11 only works up to 20 km/h
Conservatism.
The automotive industry have change cycles measured in years, if not decades. It's very hard to match up the mentality of that industry with the dynamic and unstable nature of the computer and wireless business.
Today everyone are aiming at deploying boring, basic services such as theft notification with vehicle tracking, roadside assistance, remote door unlock, crash notification etc. BMW in Europe is probably the leader where the roadside assistance call center can update a map with routes that is displayed in the new 700 series. All European solutions, BTW, uses GSM and SMS messages to transmit data. Its store and forward makes it ideal for low bandwidth mobile communication.
Next generation of services will include traffic, food, gas, service info and geo-ads. Cool things like streaming video is faaar down the road.
When we are playing around in the lab, we look at the possibility of using USB keys as storage (MP3 etc) and ignition keys.
We are also looking into the possibilities of using max-bandwidth/least-cost routing that switches between different technologies depending on what coverage you currently have. You would then have a fixed IP-address that acts as a proxy and forwards packets to the vehicle using CDPD, SMS, AMPS or whatever other links you have to the vehicle this 30-second cycle.
Things are happening, but they are not as fast moving as the rest of the Internet/telco business.
Everyone is apparently far too stupid to realize what its greatest use would be. They're too busy drooling over trying to "get a slice of the $ XX billion a year mobile phone revenue".
We've already got cell phones, and in all the places I travel for work, only once in the backwoods of Kentucky did I ever consistently lose the signal. There isn't enough "flaw" left that quality improvement in this direction is going to make much money. Not enough to justify this effort.
And as for internet access, I sure as hell don't need to share the road with some pervert whacking off to www.farmsex.com. And I don't buy the passenger angle either, the last thing parents need is for 2 brats in the back to be squabbling over which web page to visit. This just isn't very compelling.
What I want, and what I think would ultimately be useful for everyone, is a wireless net link that maxes out at about 300 yards. My car would broadcast its location (via GPS) along with everyone elses, and right up next on the dash board, I'd have a little mini-LCD with a map of the current occupants of the road. What's more, we could also send turn signals and such via this link (in addition to visually). Those places you always come to, where visual signals are ambiguous? Well, you'd now have more than just a left/right turn signal. Signaling for straight ahead, the 2nd from the left of a 4-fork road (yes, I've really seen one of thse, 5 roads meeting at a single point), etc.
Then, there is the traffic jam possibility. What if those people up ahead in a jam could warn you in time to get off at the next exit? Hell, we could even have a "thank you" and "I'm sorry" signals... might cut down on some road rage.
And when critical mass is achieved, we could start to do things that would make this even more useful. Traffic lights, for instance, would detect all the cars relevant to it. So if you're sitting at the stop light at 3am, and no one is going the other way, the traffic light would be smart enough to see this, and change the light to green for you, no waiting. Cyclic lights could die very easily... this would be very close to the smart roads they've been wanting forever.
And you know how those navigator appliances that the new rentals have, that always have the road information as it was 2 years ago? This could augment that. If a road worker plops down a orange warning cone, it starts broadcasting its location and that the road is reduced by one lane.
We could even consider getting rid of some of the traffic sign clutter... it could just be beamed directly to the dash. Instead of signs, a small transmitter mounted on the same pole, with a battery and solar cell panel. How much prettier would our roads look? Hell, you'd always know what the speed limit is (you decide if that's good or bad) because it would show on your dash. For me, I just got a ticket 2 months ago, because a road I thought was 55 for years turns out to be 45mph on one stretch. Could be useful.
And depending on how intrusive we want to let the advertizers become, we could even force them to transmit signage that way too. (Before someone gets bent, remind yourself you can turn off the mickeyD's sign on the dashboard, but we can't currently do the same with a billboard). We could concievably get rid of all signage along roads, and do so without (supposedly) crippling advertizers. Might be a bit prettier along the highways.
And why will this never happen?
#1 Idiots in Detroit like nice shiny technology, but that's as far as their understanding of it or its uses, goes.
#2 Politicians and goverment are the most worthless institution to ever exist in the 14 billion years or so of history of the universe.
#3 Some asshole would insist on making my idea more privacy intrusive than I would, and privacy advocates would go into an uproar (possibly justified).
"Besides which, cars aren't always going to be damaging to the environment. Eventually they'll be electric."
Apparently you haven't considered the fact that the electricity has to come from *somewhere*.
There are claims that humanity's electricity consumption exceeds our ability to use only minimal-impact and renewable sources of energy. Solar systems are hugely inefficent, and there is only so much wind and hydro power that can be effectively tapped.
So, things might come full circle and your electric car would actually be powered by a dirty coal plant. Some of the first experimental cars were directly coal fired.
I think that video cameras on each car might be useful. Think of how many crimes occur within view of an automobile? Many, if not most, of certain crimes (store robberies, muggings, etc).
Each person would own the camera and control its use. If there was a crime, the police could ask volunteers to check their car cams to see if they got a video record of the incident. The owner would know if they were in the area of the crime or not, either with a GPS/time correlation computer, or just by remembering.
Or, the thing could be automated. The car owner would give permission to the police ahead of time to check their camera, or on a per-incident basis. The police would query the car cam network to find out which cars were near the crime scene at the right time, and be able to fetch the video remotely.
This would be a good thing, and avoid a big brother network. The owners of the cameras are private individuals. The police would have access to the cams only by permission.
The biggest thing is that there's no reason why cam owners have to give their permission to the police. They could give permission to someone else instead, like a citizen's watch group. That way, the cams would not only be looking at the criminals, they would be looking at the police as well. Just like any crime, police abuse happens in private, and these cams would allow ordinary citizens to monitor police and look for evidence of crime.
If tits were wings it'd be flying around.
There's an angle Cringely missed on adding intelligence to vehicles. There's work going on at UC Berkeley that involves cars talking to each other and sensing where the road is. The idea is if the driver's reaction time is eliminated from driving decisions, you can pack more cars on over-burdened freeways and speed them up as well.
The way it works is there are magnets embedded in the freeway that tell a car where the road is. The cars have transmitters that communicate with the cars in the immediate vicinity so when a car speeds up or slow down, the other cars know it immediately and can react accordingly. You drive onto a freeway and pull in behind a convoy of PATH-enabled cars. The car takes over from there and drives itself until you tell it you want out of the convoy.
Instead of discouraging tailgating, the technology can use tailgating to improve overall fuel efficiency by having the trailing cars draft the leader - much like race car drivers do now except you're not relying on human reactions to make it viable. Human factors come into play as people who have ridden in a car doing 60 mph that's 4 inches behind the car in front find the experience uncomfortable.
The technology has been tested on a section of I-5 near San Diego and actually works. There are of course, reasons why it isn't going to show up in next year's models. Some are technical such as magnetizing enough freeways and dealing with magnets that go bad but a key obstacle is the need to revise liablitiy laws and draft legislation that specifies maintenance schedules and such. Without tort revision, the first accident that involved PATH-enabled cars would kill the technology. People will ignore the fact that we've had non-PATH pileups in the past and focus on "the computer did it..."