Slashdot Mirror
Automakers and Crash Data Recorders
The New York Times has a decent story about automakers not wanting to standardize car data recorders. There are a couple of nuances which the reporter mostly misses. The automakers want to avoid standardization because they can then sell access to the proprietary data format (NYT does cover this, but ignores the profit motive). The story mentions privacy issues but dismisses them as solved, yet notes that there are no privacy protections whatsoever for this data, and you can expect it to be used against you in any incident (and perhaps other times: wait until service under your warranty is refused because your car reported your bad driving habits to the dealer). That's not "solved" in my book (and I think the automakers realize that selling cars which report on their owners might backfire). Speculation about ambulance crews using crash data is just hype - no ambulance is equipped to do that, nor would I want an EMT to spend time decoding the crash data instead of, say, saving my life. The article repeatedly suggests that crash data would be used to enhance safety, without ever specifying how that is supposed to occur.
It will be an interesting debate, because on the one hand the data will be useful (enough so that the automakers think they can make money off of it) and on the other it could cost the auto owner money, either by telling the truth or lying. Yes, these boxes could err, but I would expect that to be just another factor in court. Regardless, the ramifications will be extensive.
Of course, don't forget that a tamperproof box might very well save you in court. We would all benefit from less fraud in court, as it drives up insurance. And then there are the harder to quantify but likely benefits of incorporating lots of real-world data into safety design of brakes and such.
I think it should obviously be up to the driver whether to participate. Some rental car companies might decide to use the boxes to protect their property against illegal misuse -- indeed it may be their insurers that require it; perhaps a discount could be offered to those who want to opt-out, calibrated tp the differential in insurance claims between people who use the boxes and those who do not. Monitoring is not novel: for many years I've seen speed recording devices on some long-distance buses.
All of this can be argued up or down, but I don't think a flat ban on the boxes is appropriate or likely.
On to the point. The argument that crash data may not be used is ridiculous. Most auto-development takes place on the basis of information you receive from thousands of sensors. But while we'll be seeing more and more sensors in cars for increasing safety (passive), hell, Mercedes already leads the field, we'll not be seing "developmental" sensors anytime soon, because:
1. Collecting information is one thing, applying that to meaningful solutions quite another. One must not forget that auto development is carried out by highly specialized teams, working on some pre-defined parameters. Just how would the G-forces recorded in a pile up with SUV prove useful for Mercedes new S-class? These systems (passive safety) are already incorporated before a car is launched.
2. Development requires controlled conditions. Engineers have to know all the parameters to understand the situation. This is far easily done in the develpoment stage of the product, where you can do as much testing as your budget/schedule allows.
3. Creating an "open-format" makes it easier for companies at the lower rung to access developmental data. Like it or not, this research costs money (Mercedes spent $900 million researching on their new E-class...), and companies are definitely protective of their "way". I wonder, though, just how would a Ford or a Hyundai benefit from all the data of an E-class. It will take them a few years just to analyze the data, heh.
4. Cost. Incorporating data boxes and sensors costs money. Setting up an infrastructure to make use of all the collected data costs money. While "black-boxes" are already incorporated to some extent in higher end.... I wonder how many people would be willing to fork out, say, additional $500~800 for a Honda Jazz/Fit or a Corolla that offers no additional "features" to the consumer.
5. Morality. If we can actually make sense of all the data collected in this way, would we also require manufacturers to meet some criteria in safety? Accept it, "inequality" exists. You'll be much safer in an S-class/7series at 200 Kmph then in an Corolla/Civic at 100 Kmph. Manufacturers are always trying to give more to consumers. There is good competetion already. We don't need asinine regulations.
I am sick and tired of every Tom, Jack and New York Times reporter "advising" auto-manufacturers how to develop cars. Just shut the fuck up. Perhaps these retards don't realize, but modern automoblies are incredible feats of engineering. Thousands of components work together to ensure that some epsilon minus moron can reach his/her NYTimes office and conjure idiotic articles.
Beleive me, automobiles engineers/manufacturers are far more informed in this matter. This is not 1960s. They are already doing an absolutely terrific job, and can do well without such retarded ideas on important matters of safety.
I work on devices similiar to this under a research grant. Ours having nothing to do with the auto manufacturer. I am also limited in what I can say in regards to the technology to implement it, due to NDAs (I could post as AC, but since I'm the ONLY programmer, it would sort of be obvious anyway).
There are a number of uses for the data, beyond litigation. For instance, while there are plenty of numbers of how your Dodge Neon survives running in a brick wall at 20MPH, there's little real world data on the stresses placed on your Dodge Neon when it gets clipped at a 45 degree angle in the right rear quarter by a UPS truck. The onboard accellerometers record this information a certain amount of time before and after the impact event and upload it to a server.
In theory, the data can be used to determine the level of EMS dispatch. If your vehicle is sitting upside down on it's roof, it's far more likely that there are injuries than if you took a 12MPH hit into the rear bumper. Using cellular technology, a voice channel can be opened to the occupant to ask "Are you dead yet?". Or perhaps as EMS is being dispatched anyway, the driver can say "No, everyone is fine. We don't need EMS." This saves money and allows EMS to be dispatched to people how NEED them. And at rush hour in major cities like Atlanta, there are a LOT of accidents, particularly in inclement weather.
Road speed limits are calculated by people who supposedly know what they're doing. If you're like me, you probably wonder how such a title as "Traffic Engineer" exists, since there seems to be little evidence of any "engineering" at some intersections and traffic lights. Data from boxes like (real time data, not just crash data) that show a drivers average speed through areas can be used to calculate how traffic DOES flow, not how it SHOULD flow.
The research boxes we place don't give data back to the volunteers. However, if you're a parent, and have a dependent driver on covered by your insurance, shouldn't and don't you have a right and responsibility to know how that dependent is driving? If Jr. is taking the family SUV up to 126MPH on the local express way, don't you want to pull his keys, if for NO OTHER REASON, to keep YOUR rates low?
Such data can also be used to enforce cost-per-mile insurance. Suppose we both drive 5 miles to work, but I take back roads and you take expressways. Statistically, one will be more safe than the other, depending on area (for instance, back roads are probably LESS safe in mountainous terrain). Why should we pay the same insurance rates when one of us is driving a statistically safer route? And this data can also be used to catch people who lie to the insurance companies and say "Oh yea, I only drive 3 miles to work." when they're actually driving 30.
Emissions information can be gleaned from this data also. When emissions and milage are generated for a given model, it's based on a professional driver, driving a fixed course. The professional driver is needed so that each car is treated the same. However, the driver knows how to drive to meet the requirements. So while uniform, they're loaded numbers. The EPA would like real numbers from real cars driven by real people (for instance, my car has two throttle positions: wide-ass open, and idle). While many view the EPA as an intrusion, they also serve a useful purpose. Numbers for the Koyoto (sp?) Protocol are based on calculated emissions, not measured. I think we can all agree that breathing clearing air isn't going to hurt us, and if we're going to have numbers for how many tonnes of carbin dioxide are spewed into the air each year, they may as well be actual. Side point: Did you know that a standard two cycle engine JetSki spews more pollution in 45 minutes than any American made car manufactured in the last 5 years does in 100,000 miles?
Now, there is a dark side to all this techology. Our boxes are placed in volunteer vehicles, and not installed by manufacturers. There is a few sheets of legalise that people sign (that I haven't read) that indicate what this data from this device can and cannot be used for. I am TOTALLY against the anyone knowing where I am at all times, and how I drive. I believe in anonymous data, but I also know that data is rarely stays anonymous. It can be cross correlated with other databases, and allow people to figure out things I really don't think they have any business knowing. I have no idea how to handle that particular issue.
Lawyers and insurance companies would LOVE devices like this installed. Hell, they'd like to have cameras, pulse rate monitors, blood alcohol monitors, and orbital satellite laser platforms. Their reason is a little valid: Make the culprit, not the victim, pay. But since insurance is nothing more than government sponsored Mafia gambling, you know that will be abused in any way possible. And used to weasle out paying the victim, one way or another. (yes, IANAL, and if I had any offspring that became lawyers, I'd shoot the little bastard.)
There IS a lot of useful and cool data that can come out of this. It is a step to drive-by-wire cars, which if it eliminated accidents on expressways, I wouldn't be adverse to enabling. But there are a lot of ways this data can be misused, and I'm confident that Congress and the insurance criminals will figure out a way to abuse it.
--jcwren
Holding car drivers to the same standards as aircraft is such a huge leap that the paperwork generated by it could likely employ everyone in America.
While you'll probably get a lot of "not practical/desirable" responses to this, I think you have touched upon a valid point: if these devices do end up making it in every vehicle, understanding exactly what data is recorded and what is considered to be within desired tolerances by manufacturers, insurance companies and law enforcement agencies should not only be made public, but also be part of written driver's license exams.
I can see it now when driver X gets rear-ended by a drunk on a cell phone and the drunk's insurance company downloads driver X's black box data and finds that driver X's radio volume was 0.001dB above what they considered to be acceptable and therefore they deny all claims. Of course, the operator had no way of knowing this ahead of time because it was proprietary information; the drunk walks and the insurance company finds another way to fuck the public.
moto411.com