Statement of Mike Horne, AD Aerospace

MR. HORNE: I think I'm the only person in three days that's had his full biography read out.

Good morning, ladies and gentlemen. I'm delighted to be here to talk to you about our vision of future accident recorders across all modes of transport. I've noticed over the last few days that various speakers have highlighted the differences between modes of transport. I think this symposium has emphasized without any doubt that the similarities far outweigh those differences.

When an incident or accident occurs, what do the investigators need to wholly explain and understand the circumstances? They need witness statements from the involved parties and from bystanders; they need to examine the accident scene and see what they can get from that; they need to have any vehicle instruments that may be stuck during the accident to give them further data; they need any recordings of the environment at the time of the accident; weather reports; and so on.

There have been advances in all the modes of transport. We're talking now about Voyage Data Recorders, Automobile Black-Boxes, Railroad Event Recorders, but air transportation still really gives us the lead and direction.

What are the problems with existing recorders? There are problems with even existing air accident recorders. The first one is that there's too little information. Thirty minute audio is currently being increased to two hours and we have talked elsewhere about the parameter list being raised to 88 and beyond. There's not enough detail. A lot of the CVR recordings are of the form of: "Why is that doing that?", and "Is that supposed to be like that?", which doesn't really advance the investigation very much.

And we have data being lost or not recorded at all. When aircraft power fails, then the recording fails; high intensity impacts, and fire damage the recordings; and we are being overtaken by new technology, for instance digital displays eliminate jammed instruments, and digital messaging to the pilot replaces speech.

What of the Future Direction? In the future, the accident recorder will be digital, which gives us high shock resistance, high temperature resistance and therefore, higher reliability. Improved survivability of accident recorders as signified by the advent of two hour fire testing, and increased testing requirements. The Future Recorder power supply will be independent of the vehicle, and there'll be video recording. We've heard quite a few different inputs on video being used in accident recording, and I think we're going to hear some more this morning. The only way really to get all the available information around the environment at the time is video recording, which is what I want to concentrate on.

First. a slight diversion to illustrate what even live cameras can do to prevent accidents. Let's consider the benefits of improved vision. In an accident in 1991, an aircraft had a tire burst into flame on takeoff roll. Not aware of the fire, the undercarriage was retracted as normal into the fuselage. The total loss of the aircraft led to 261 casualties.

In a similar incident, a wheel well fire caused a fire warning soon after takeoff, but an aircraft following out warned the pilot by radio that they could see flames. The aircraft returned to land as an emergency, but despite extensive damage to the aircraft, there were no casualties. This shows the immediate saving opportunities of giving a clear view to the pilot of the outside of his aircraft.

How is video used in society today? The answer is "very widely, indeed!". Video cameras record all our most precious moments, even births, I'm told! In city centers they're widely used to deter violence by allowing people to be followed and apprehended; banks and stores check on credit cards and ATM fraud by logging the card user at the time of the use; and the police find various uses of video for logging speeding offenses, a practice which I now believe is being introduced into the U.S.

What about video in the workplace? It is still fairly controversial but gaining support as giving a true record of what actually happened. Stores use it to catch and prosecute thieves; banks, to protect employees against theft; video is used in police cars, both to prevent assaults on the police and to protect citizens from overzealous policing; and school buses are using video now to prevent violence between pupils, to prevent vandalism to the school buses, and to prevent litigation aimed at the school transportation authorities.

As you can see in these cases, video is largely used to protect the individuals being recorded. So, is video recording in the workplace acceptable or are the issues of privacy, which we've heard so much about, insurmountable? That rather depends on the relationship between the management and the employees. Is it trust or suspicion? Ideally, videos should be used for the protection of both management and employee and for the persecution of neither. In any case, it is vital in respect of that trust that recordings are erased, once it is established that no incident has occurred and no investigation is required.

Let's return to the specific of the airline cockpit. What are we looking to record in the cockpit that will really help the investigative process? I believe that we want to view the pilot, which may be a little controversial, but there you are, we need to know who is in control, and whether they are being hindered or impeded by anything? We need to monitor the controls, are they being obscured by anything extraneous, a smoke filled cockpit, or papers falling across the instrument panel? We need to check the instruments for functionality and whether the graphics are correct, but we're not trying to read text or data from instrumentation, that can be done better other ways. This is a video clip from the cockpit of the 737 [Video Playing]. Note the amount of nonverbal communication going on.

It important to know absolutely who is flying the plane at the given time; were there extraneous circumstances or a distraction; and whether his actions are correct. The buzzword is Cockpit Resource Management, and it is vital to understand in the circumstances of an incident or accident.

Technically, we need a lot of memory. An uncompressed video frame is around 200 kilobytes per picture. With compression techniques, we can get that down to 15 kilobytes. But this still means that at two pictures per second from a single camera we get an hour of recording per gigabyte, or 24 hours in the largest available hard drive. This can be improved, of course, by using methods like video motion detection so when there's movement in the picture, only then will the images be stored; or by recordings triggered by alarms or events.

There are several possible video compression techniques. JPEG is currently the preferred method as it allows each picture to be wholly compressed into a data file, whereas MPEG derives frames from a sequence, storing mainly the differences and use interpolated frames which really don't stand up in court. Other techniques, such as Fractals and Wavelets are possible, and evaluation of their benefits is being addressed.

So the future accident recorder will consist of three parts: data, video and audio sections, and the total amount of crash protected memory needed will be around one gigabyte, as compared to the current requirement for about 60 to 80 megabytes. Quite a step! Even that will only give 25 hours of data, two hours of audio and 30 minutes of video, which is about the minimum that we can look at.

We should consider the recorder as a "Data Sponge", absorbing whatever data is given to it, be it digital data, digitized audio or digitized video, with the recorder merely recording that data.

All the digitization can be generated locally to the sensor, simplifying the recorder itself. That does make the replay slightly more complicated, with more involvement needed from industry to separate out the various portions.

All of this leads to the concept of the Accident Network Server, using digital transmission rates at 10 to 100 megabytes per second, standard Ethernet technology allowing transmission of data by satellite, ISDN telephone line or any other way that standard office communications can be done. This means that you also are able to use a standard Internet type browser technique to review and control the network server and simple user friendly commands using TCP-IP and HTML pages, very similar to the Internet.

This gives us a multi-modal solution. While all the examples here are taken from aircraft, the technology is applicable across all transportation modes, and is currently being evaluated for use on UK trains, proposed into U.S. school buses, as well as UK airlines.

So what is the current state of the art? I have put this slide in to show that the technology exists now and is available now from AD Aerospace, among others. The FV-0730 is an aircraft video server using spread spectrum transmission from the aircraft to a terminal building, and both internal and external cameras. This is mainly used as a security aid at the moment.

The TransVu 0745 is a ground vehicle video server, with touch screen control panel. The FV-0700 crash proof version is under development to combine video, audio and data recording, as we've seen.

Looking at other recorder technologies, Military and FOQA capabilities far exceed the current accident recording requirements. Various manufacturers are producing crash protected digital recorders in line with current requirements; 25 hours data and increased perimeter lists, and two hours audio, but companies like Terrington Systems in the UK currently record 15,000 parameters with all data collected, all data time stamped and recorded to that tape, allowing a full simulation of an incident. Maybe that's the way we should be going.

In summary, then, the future multi-mode accident recorder will be applicable across all transportation modes; it will be digital; it will be independent of vehicle supplies; it will be high integrity, and high capacity giving more than 24 hours recording and it will be capable of data, audio and video.

This picture is from the Kegworth crash in 1989, and shows what we're trying to avoid.

Further information about the video recording as an accident investigation means is available in my paper, which is in the proceedings of the symposium, which also covers in more detail the mathematics behind the JPEG compression method and the comparison with MPEG.

That's about what I have.

Thank you very much for listening.