Statement of John Hinch, NHTSA¹

Mr. HINCH: Thank you.

[S-1] Good morning. May I see, with a show of hands, how many of you have driven or ridden in a car or a bus in the last week or two? Probably most of you; right? So I think I'm in the right place, and you probably will have some interest in what I have to say today.

On behalf of my co-authors, Tom and Keith of GM and Chip from my office, and also being the proud owner of a very short straw, I am here to represent the four of us in this presentation.

[S-2] Crashes are an every day occurrence in America. I reflect back two days to Chairman Hall, when he said that in the transportation industry, 95 percent of the harm, 95 percent of the fatalities, come from driving our automobiles and trucks on the nation's highway every day.

I also reflect that since he made that speech just two days ago, 250 people have died on the highways. About another 20,000 people have been injured in over 50,000 police reported crashes and probably another 50,000 more minor crashes that weren't reported. The opportunities for collecting crash data on personal automobiles is immense.

[S-3] There's been interest in collecting crash data on automobiles for a long time. In this 1914 crash in Scotland, you can see the quizzical look on these investigators' faces trying to figure out what caused this purported first crash in Scotland.

[S-4] This problem carries on to America. In an early multi-vehicle crash you can see investigators talking to each other. "Did you get the readout from that event data recorder, Joe?" "No. The batteries are dead on my PC. Let me get a power cord. We'll get the right out."

[S-5] As I said, the opportunities are vast-eighteen thousand tow away crashes a day. If we were to buy those crashes as a society, that is, conduct 18,000 crash tests, we would have to spend $600 million to run an equivalent number of crash tests as there are in one day's tow away crashes. And there are more crashes beyond that. That's nearly double NHTSA's annual budget. That's a huge investment America makes every day.

The current output of crash tests that are conducted for safety purposes by industry and government for U.S. produced vehicles is about 5,000 a year. Not nearly as much as what's conducted by the citizens every day.

And, in that two-day period I discussed, since Chairman Hall spoke, the cost to America of those crashes has been $1 billion. That's what we have spent as a society. In one year, we spent over $150 billion as a society for automobile crashes, truck crashes, heavy truck and bus crashes.

[S-6,7] Well, I'm here today to tell you about an exciting new field of crash recorders. This work was started as a NHTSA need for crash data. We received, just a couple of years ago, an NTSB recommendation to get busy in this area. NTSB suggested that we develop and implement, in conjunction with manufacturers, a plan to gather better crash data.

[S-8] About that same time, NHTSA contracted with one of its federal partners, NASA, to conduct a review of advanced air bag technology. The Jet Propulsion Laboratory recommended, to NHTSA, that crash recorders already existed on many manufactured vehicles on the road today and that we should set up a program to collect these recorded crash data.

[S-9] JPL also recommended that NHTSA also look at other uses of these data, such as emergency rescues. And, as the NTSB said, we should look into the questions that arise related to who owns the data, legal issues, et cetera.

[S-10] There are many uses for event data recorders. Being a scientist at NHTSA, my interest is in improving vehicle design and highway safety. As I mentioned earlier, the challenge is great. At NHTSA, we think, that this is one of the next great tools that will be available to our researchers to improve safety.

We can also use these data for improving information to set better regulatory requirements and to produce consumer recommendations. These data could be used for crash reconstructions for both NHTSA and other reconstructionists such as state police or insurance investigators. It could also be used to develop driver information from pre-impact crash information, et cetera.

[S-11] At NHTSA, we look at crashes using the Haddon Matrix. The Haddon Matrix is named after Dr. Haddon, NHTSA's first administrator who divided a crash into nine components. He looked at the pre-crash, the crash, and the post-crash for the human, the vehicle, and the environment.

Without EDRs, when we look at a crash, we can get a lot of information about the post-crash but we cannot get very much information about the crash nor the pre-crash.

[S-12] But with a visionary EDR of the future, we can find a lot about the crash. We can find information about the human; what (s)he was doing before the crash. We can find out information about the vehicle; how fast it was going, were the brakes applied, was (s)he using ABS. We can even find out information about the environment; was it raining, was it icy, was it windy. All this information would help us understand why all those crashes occur and help us move into the future and reducing crashes.

I think I remember several of our speakers saying during this convention that crash rates have helped decrease crash frequency in other modes, maybe 60 or 70 percent in the one slide shown for aviation. If we could gain a one-percent reduction with these devices, just one percent, that would save 400 people's lives a year. So the gains here are immense!

[S-13] Today, I would like to talk, on behalf of my co-authors, about an exciting system that's now available and has been available for several years on many General Motors' light cars and trucks.

These motor vehicles have had air bags for several years. For an air bag system to make a decision about whether it should inflate the air bag or not, it has to analyze the crash. So all air bag equipped cars have crash analyzers.

A crash analyzer typically uses crash deceleration and other pertinent data, so those data related to the crash already exist in the crash analyzer. It's a simple step to record those data at the final step of deploying the air bag.

[S-14] General Motors started recording data with its1990 Model Year air bag equipped vehicles. In 1994, they improved the air bag control module and it included additional parameters that were recorded. And in 1999, they have improved that air bag control module one more time to include some pre-impact crash data.

[S-15] Looking at a simplified block diagram of the 1999 air bag control module, which includes the event data recorder, we can see that the recorder collects engine speed, vehicle speed, brake status, whether it's on or off, and throttle position. These data have already been converted to digital values and are transmitted through the car on a serial bus that connects many microprocessors and components in the car together. The data are collected off the bus and processed by the SDM and then stored in non-erasable memory.

[S-16] One important item the EDR records is information on the crash pulse. In the 1999 EDR model, it computes the cumulative change in velocity and accumulates that in a register. Then each 10 milliseconds, as the processor accumulates, it reports out the current delta-V value which is stored.

Here's a typical delta-V trace, from a crash, where the x-axis is time in milliseconds. As most of you know, automobile crashes typically last less than 100 milliseconds. That's a very short time-one tenth of a second-for the entire crash.

During the first part of the crash, the crash analyzer has to decide if a crash is occurring and then deploy the air bag, if it is required. Crash pulse data will help manufacturers and regulators to understand the crash related characteristics, such as, crash forces and crash durations.

[S-17] This chart shows typical pre-impact data which were recorder with a 1999 EDR. And let me point out for a second that in the General Motors fleet, as other manufacturers, these devices aren't all incorporated in the same year. They're phased in over several years. The1999 EDR is being phased in with some models in 1999 and should be completely phased into the GM fleet over the next few years.

Here we see four channels of information that was recorded. It was recorded five times, once each second prior to the crash.

In this case, we see a vehicle moving along at some speed with the brake off, and then at three seconds, (s)he switches from brake off to brake on. And you can see the engine speed, vehicle speed, throttle position all decrease with time.

[S-18] General Motors has reviewed the accuracy and resolution of these different parameters; and more on that can be found in our paper.

[S-19] General Motors and NHTSA have a joint program to look at how well the EDR predicts crash data. So we took GM EDRs from NHTSA's NCAP test vehicles. For those of you who aren't familiar with NCAP, it's the New Car Assessment Program, the program where NHTSA conducts crash tests at 35 miles an hour into a flat barrier with the front two occupants fully belted.

During those tests, we collect high fidelity electronic acceleration data. The acceleration data were used to compute delta-V, which was then compared that to the output of the event data recorder. You can see here a comparison of those two.

[S-20] Well, one might ask-this is great that GM has collected all this data. What can be done with it? It's probably in one's and zero's located in some box that you'll never be able to find inside your car. And you are correct. It's probably hard to find. But they do have a processes that can be used to collect the data from the vehicle.

In the past, the process for doing that has been to use a GM service tool, the GM Tech-1 service tool pictured here. But, GM is about to announce for sale a new tool that will be available to the public. It's going to be manufactured by a company GM has licensed to build this product. I believe the tool will be available sometime this summer.

[S-21] The name of the company is VETRONIX and the tool is pictured here. It's going to be a set of cables and computer software. Using a laptop computer, the software will allow you to interrogate the event data recorder and collect the data.

If you have a 1999 model EDR, you would be able to collect the pre-crash data, plus a host of other crash related data. If you have a 1994 type EDR, you would be able to collect delta-V as well as other types of information.

[S-22] Well, one might ask what can be done with these EDR data. Here you can see, the GM guys and myself, trying to figure out exactly what to do with these data. Well, I have to admit on that day we never did figure it out.

[S-23] At NHTSA, we have been using the EDR data for about two years. We have been working in conjunction with GM where we will take a box to GM for data retrieval or we'll have a GM representative come to our crash scene and collect the data. Here's a sample of about 10 special crash investigations that involved General Motors' cars that were equipped with the event data recorders.

This shows a comparison of two data elements that were of interest. One is a comparison of driver belt status, comparing what the crash investigators observed to the event data recorder. We also compared the delta-V that the crash investigator computed, using a computer program, that's based on evidence obtained at the scene to the EDR data.

[S-24] One specific case, I'll spend a second talking about, was a car impacting the back of a large truck. The car under-rode the back of a large tractor trailer, resulting in a severe under-ride crash. To estimate the delta-V for vehicles in a crash, NHTSA investigators use a computer program, called WINSMASH, which uses post-crash evidence for input parameters. WINSMASH requires the investigator to enter stiffness coefficients to define the frontal characteristics of the vehicle, which are often based on full-frontal barrier crashes. In unusual crashes, such as this one, the frontal characteristics used in the program may not represent the actual characteristic, since only a portion of the vehicle was involved in the under-ride crash. This could result in the predicted delta-V being incorrect.

For this vehicle, WINSMASH produced a delta-V of only 23 mph, but the investigator felt the delta-V was much higher. Decoding the data from the EDR indicated the delta-V was closer to 50 mph, which the investigator felt was closer to the actual than the WINSMASH program.

Also in this case it was not clear to the investigator in their early parts of the investigation whether the seatbelt was worn. The witness marks normally associated with seatbelts just weren't there. By using the EDR, the investigator was able to determine whether or not the belt was used. In this case, the belt was not used.

[S-25] Well, at NHTSA-we heard the FRA² people talk about their advisory committee- we have an advisory committee for motor vehicle safety, called the Motor Vehicle Safety Research Advisory Committee. It's made up of industry representatives, researchers, universities, and representatives from the medical field. In 1998, NHTSA staff presented a briefing to the MVSRAC committee members, recommending that a working group be formed to work on event data recorders.

[S-26] The MVSRAC members agreed that NHTSA should form the working group but did indicate there were some limitations and that for some manufacturers it would be several years before event data would be available. The MVSRAC working group was formed under the Crashworthiness Subcommittee.

[S-27] The working group has representatives from manufacturers, government, private citizens, universities, and the medical community-and includes representivies from Transport Canada and the NTSB.

[S-28] The objectives of the MVSRAC working group are to define functional and performance requirements for onboard recorders, understand the technology presently available in event data recorders, and develop a set of data definitions. With several different users of EDR data-crash investigators, government users, manufacturers-there are needs for different types of event data. This has produced a lot of interplay in this group trying to figure out which are the important data and the uses of those data.

[S-29] Another interesting issue, remembering back to the JPL and NTSB recommendations, are the privacy and legal issues. Some major issues include: who owns the EDR data and how the data are used. We have a lawyer on our panel to help us in that area.

Another objective is the historical overview of NHTSA's actions related to data collection, and the inclusion of these data in the FARS and NASS-CDS databases.

[S-30] Potential outcomes; and I say potential because this is a working group made up of a committee and I'm only one of many, so it really will be a group output when we get done. We do have some idea of putting out a technical report. And I would expect that report to be done in about a year and a half.

Under the MVSRAC, the purpose of a working group is to collect information, like a fact finding group. After we complete this task, we will make recommendations to the full committee. It will be up to the committee to give guidance to NHTSA on what should be done next. We would think that the committee would encourage NHTSA, based on what we've learned to far, to establish or use existing national databases for EDR data and encourage other manufacturers besides General Motors to develop event data recorders.

[S-31] Some conclusions. It is my feeling that event data recorders clearly have the potential to reduce the fatalities and injuries on the nation's highway. I truly believe that. I also believe, based on this working group, that a well coordinated effort will be needed for NHTSA and industry to meet the recommendations of the Safety Board.

NHTSA's MVSRAC event data recorder working group will establish some guidelines which will be available to all manufacturers and users of event data in the future. Event data are now being stored. As I mentioned earlier, we have nine or 10 cases, and more are in the pipeline for analysis. We have modified our NASS-CDS data file (the National Automotive Sampling System - Crashworthiness Data System), which is a national estimate of crashes in the United States, our special crash investigation file and other databases so they have holders to collect these data.

[S-32] So to end, if you have time during or after the break and you want to see what an event data recorder actually looks like, you can stick your head under the hood of a GM vehicle at the VETRONIX's poster site outside the meeting room. General Motors was gracious enough to bring some people from California who brought one of these VETRONIX tools and you can take a look at how a prototype or "beta" version of one of those tools would work today and possibly get a vision of what they would look like a couple of months from now.

Thank you very much.

1 NHTSA-National Highway Traffic Safety Administration

2 FRA - Federal Railroad Administration