Thank you, Gary, for that kind introduction, and thank you for inviting me to speak today on behalf of the National Transportation Safety Board. As the only attorney on the Board, I appreciate and enjoy opportunities to reach out to the legal community.
Many of you are very familiar with what the NTSB does. We are the independent federal agency that is charged with investigating transportation accidents, finding their causes, and making safety recommendations to prevent recurrences. We cannot require recipients to implement our recommendations, so it is a testament to the quality of the work by our amazing staff that more than 80% of our recommendations are nonetheless acted upon favorably.
For those of you who may not be so familiar with what we do, I would like to note that we are again offering our class, “NTSB Investigations: What Legal Professionals Need to Know” at the NTSB Training Center in Ashburn, Virginia, on March 29-30. This course, which explains what to expect if one of your clients is designated to be a party to an NTSB investigation, is usually well attended.
Today I would like to talk about the effects of information technology advances on our accident investigations, how we have addressed new technologies that we encountered in our investigations, and NTSB’s ongoing vigilance to the challenges posed by evolving technologies, including cybersecurity.
IT improvements are helping us conduct more effective and efficient accident investigations in several ways. First, we can obtain data from many IT systems on the vehicles themselves. Aviation is the most advanced in this regard because they have had “black boxes,” i.e., crash-hardened flight data recorders and cockpit voice recorders – which, of course, are actually orange – for decades. More recently, event recorders are also being installed in trucks, trains, and ships.
In Part 135 operations, where not all aircraft are required to have crash-hardened recorders, we have recommended that the FAA require operators to install devices capable of supporting flight data management programs. These programs use data to spot trends with adverse safety implications, similar to Flight Operational Quality Assurance programs in the Part 121 airlines.
We have also been recommending on-board cameras for many years, in some cases both outward facing and inward facing, and some have implemented our recommendations. For example, the railroads are realizing that they often benefit from an outward facing camera that shows cars going around the barriers at a grade crossing or that a person who was struck by a train was committing suicide. Inward facing cameras have been a more difficult battle, but as I will discuss later, we have found that even with the sophistication of existing recorders in aviation, inward facing cameras can provide valuable information about the actions of the crew that is not otherwise available.
In addition to recorders that were designed for accident investigation or flight data monitoring purposes, we are also enjoying the benefits of a proliferation of other on-vehicle systems that acquire and process data. To the extent that this information is (a) recorded, (b) not volatile, i.e., not lost when the power is removed, and (c) recoverable despite damage to the memory chip, these on-vehicle systems can add significantly to our investigators’ ability to learn facts about an accident.
For example, the computer chips that control the engines, antilock brakes, and airbags in motor vehicles often provide information such as speeds, time of brake application, and impact force in our highway safety investigations. As driver-assistance technology becomes more common, and potentially contributes to safer roads, cameras and other sensors will also continuously acquire data. In general aviation, the increasing incorporation of glass cockpits is providing additional information for our investigations.
Having said that, I should add that most of these newer operational data sources were not designed for the purpose of informing accident investigations or for data monitoring, so as we go down that path by increasingly using the data for those purposes, we are early on the learning curve of other issues that are raised, such as privacy and liability issues. Stay tuned as we explore these new realms.
We are also enjoying the benefits of an increasing number of sources of accident investigation information that are not on the vehicle. The growth of surveillance cameras has been exponential, and they are used for a variety of reasons, such as to help solve crimes. In aviation, accidents and incidents at or near airports are often captured on the airport surveillance cameras. In other modes, we often benefit from surveillance cameras that were fortuitously pointed toward an accident site. For example, our investigation of the collapse of a bridge over the Skagit River in Washington State in 2013 was aided by images from a nearby surveillance camera that captured images of the oversize truck hitting the bridge structure and the subsequent collapse of the bridge.
Another increasingly important source of video that has been very helpful in our accident investigations is cameras owned by individuals, such as in their personal phones or dash cams mounted in vehicles. These sources often provide valuable information that is not otherwise available. For example, when a Boeing 747 cargo jet crashed in Bagram, Afghanistan. In 2013, the information from the flight data and cockpit voice recorders stopped when the aircraft rotated to take off because when it rotated, the 12-ton vehicle in the rear of the aircraft broke free from its tie-down straps and struck the aft cargo bulkhead. The impact broke the connections to both recorders, which were attached to that bulkhead, after which the recorders stopped collecting data. Luckily a truck that happened to be passing by on a nearby highway caught the entire event – takeoff, climb, stall, and crash – on its dash cam. That dash cam, plus the airport surveillance cameras, provided valuable information that would not otherwise have been available.
Similarly, we are benefitting more from video images of crashes from individual phone cameras that people voluntarily provide to us. In 2015 it was estimated that there were 2.6 billion smart phone subscriptions in the world, expected to grow to 6.1 billion by 2020. The NTSB is seeing, and using, the resulting photographic data in our investigations. Spectator imagery is not new; rather, the ubiquity of this data source is.
For example, when a P-51 Mustang crashed into a crowd of spectators at an air race in Reno, NV, in 2011, killing the pilot and 10 people on the ground, we obtained extensive video and high-resolution photographs from the spectators. Those video sources enabled our investigators to determine that, when a piece of the elevator trim tab separated from the airplane, the airplane underwent a 17-g pull-up, rendering the pilot unconscious, whereupon the airplane crashed. That information would not have been available from any other source because few g-meters go above about 9 g’s. Another example is the Asiana crash in San Francisco, CA, in 2013, in which the post-impact cartwheeling of the airplane was caught on camera by a bystander. Because the tail separated on impact, that post-impact motion did not register on the flight data recorder.
Unless these public video sources are posted on social media, the only way we become aware of their existence is when they are provided to us voluntarily. That is one of the reasons that our reputation for fairness is so important, because if people do not believe in our mission and how well we accomplish it, they might be less willing to come forward with their information. In that regard, it is unfortunate that the news and popular media sometimes portrays the NTSB as being adversarial and confrontational. Not only are we concerned that this might make people who are involved in accidents less willing to cooperate with us, but also that members of the public who have helpful information about accidents may be less willing to provide it to us.
One more area in which our investigations benefit from the latest IT advances is our own use of such technologies. Although there are many examples of such technologies being used in our research and engineering laboratories, one example with field applications includes our use of unmanned aerial systems, commonly known as drones.
Drones can be flown over accident scenes to gather data, including accident scenes where humans should not or cannot go. In addition, drones can be combined with other IT advances such as laser scanning, which the NTSB uses to get precise measurements of accident vehicles and accident scenes in 3D. We can use flyover documentation from the drone in combination with ground-based laser scanning or use the drone as a platform for aerial lidar. From there, items of interest can be 3D-printed and compared with an exemplary item, which might itself be printed from a blueprint. The same data can be used in accident sequence animations to help the investigator – and ultimately the public – visualize the accident.
We now have multiple FAA-certified drone pilots on staff, and we regularly use a drone to document accident scenes. For example, we used a drone to help document the scene of the recent train accident in Hoboken, NJ, that you might have seen on the news.
These IT advances can, however, also introduce challenges. As you know, we rigorously control information that is provided to the media about an accident investigation while the investigation is underway. We do that because our investigations depend heavily upon willing and meaningful engagement by the parties, and the parties are more willing to engage if they feel that the process is fair. One of the ways that we help ensure fairness is by being very careful that information regarding an investigation that goes to the media is in proper context and not misleading. When the public posts information about an accident on social media, we are less able to ensure that the parties view the investigation process as fair.
That’s a brief description of some of the ways that IT advances affect our investigations. Now I would like to discuss how we respond to new technologies when we encounter them in our investigations. It’s my pleasure to lead an agency of investigators with some of the best minds in their respective fields. They come from a variety of backgrounds, and they are passionate and highly motivated about improving safety. They continually seek out research in their fields, as well as training opportunities to sharpen their skills and expand their perspectives. More specifically, in order to “keep up” with the rapid technological advances that we are seeing in all of the modes, staff aggressively pursues advanced training; reaches out to visit and tour locations where those technologies are being used; and draws heavily upon the parties to our investigations, as well as other outside experts, for their technical assistance, expertise, and peer review.
An example of our investigation of new technologies involved the lithium ion battery thermal runaway on a Boeing 787 at Logan Airport in 2013. At the time of the event, we had little expertise or experience regarding lithium ion batteries. GS Yuasa, the supplier of the 787 battery, was a source of invaluable technical assistance throughout the investigation. We visited other battery manufacturers to learn more about the manufacturing process. We held a public forum on lithium batteries in transportation, in order to broadly capture the variety of battery chemistries and potential for failure, based upon experiences in various industries and the military. Then we held a public hearing on the incident, capturing information on the record in support of our investigation. Our final report reflects how thoroughly the staff went about acquiring the expertise necessary to be confident in our determination of probable cause and in our recommendations to prevent a recurrence.
Another example relates to our investigation of two test flight accidents. I refer to test flight accidents as the “airliner accident of the future” in which the telemetry of data to a ground station will be so extensive that it will eventually – in the distant but foreseeable future – eliminate the need to obtain information from the recorders. It is just a matter of time before airliners will be sending all of their information to a ground station, either directly or via satellite. On one hand, that is good because it means we will no longer have to search for the recorders, such as with Air France Flight 447 that crashed in the Atlantic Ocean in 2009 or the loss of Malaysia Flight 370 in 2014 that has yet to be found. That means that our investigations will be more focused and concluded more quickly, and will not risk losing data due to inability to find the recorders. On the other hand, we will no longer be using our present procedure of taking recorders to our lab in Washington and reading them out under carefully controlled circumstances, so we will have much less control over the flow of information.
One such test flight accident was the test launch in 2014 by Virgin Galactic of SpaceShipTwo, the rocket-powered vehicle made by Scaled Composites that was designed to take six paying participants to an altitude of 63 miles to experience near-space travel and to see how our planet looks from that altitude. Among the extensive telemetry from that test flight to the ground stations was the image from a ceiling mounted camera that showed what was happening in the cockpit.
Based upon the images from that camera, we were able to announce to the public on the second day of the investigation that the co-pilot moved a lever at an inappropriate time, which resulted in the disintegration of the space vehicle. That information, which was considerably more than we could have obtained from a cockpit voice recorder and a flight data recorder, enabled us to complete a much more focused, effective, and efficient investigation in only nine months, which is record time for an investigation of that complexity. However, it also gave us very little control over the dissemination of information about the co-pilot’s actions because the images from the camera were sent to several control rooms in real time and seen by many people who had no obligation of confidentiality. Transparency to the public is in our DNA, so the public will know the basis of our conclusions and recommendations, but our public announcement of the crew’s actions that early in the investigation was unprecedented.
One very challenging task in the SpaceShipTwo investigation was the need to “synch up” different streams of data which had slight time offsets. This was accomplished by matching key events in the data. Another challenge that emerged was the preparation of video transcripts to document in written form what the camera captured – a process that differs greatly from the preparation of an audio transcript.
By way of a footnote to our Virgin Galactic investigation, our investigators apparently earned the respect of others in the spaceflight industry because when a SpaceX test-launch exploded on the launch pad earlier this year, SpaceX invited NTSB to investigate even though the accident does not meet our criteria for investigation. We are very honored that this invitation demonstrated the commercial space industry’s respect for the integrity of our accident investigation process.
Another test flight accident that we investigated was the crash of a Gulfstream 650 in Roswell, NM, in 2011. Again there was extensive telemetry to the ground station, which provided more parameters at higher sampling rates than was available from the flight data recorder. This was the first time we had encountered such a situation – telemetered information that was more extensive than the information on the flight data recorder that we sent to our labs in Washington – and it was a major learning opportunity for our staff.
A third example of our investigation of new technologies is our investigation of drone accidents. The first drone accident that we investigated was the crash of a Customs and Border Protection drone near Nogales, NM, in 2006. Since that time we have investigated more than 30 accidents and incidents involving drones. Fortunately, all of our drone accident investigations have been single vehicle accidents, and we share the widespread concern that a drone may someday collide with a helicopter or airplane.
A final topic I would like to touch on briefly is cyber security, which involves two aspects. The first aspect, intentional interference with digital systems, is not in our lane because we do not investigate criminal wrongdoing. After the crash of TWA 800, the FBI and the NTSB entered into a Memorandum of Understanding that if we find evidence of criminal wrongdoing, we will turn the investigation over to the FBI, whereupon the FBI will lead the investigation and we will provide technical support. We have a longstanding productive relationship with the FBI, and even when there is no criminal wrongdoing, they offer extensive assistance to many of our investigations.
The second aspect of cyber security, which is in our lane, relates to the fact that both the digital systems and the cyber protections for those systems are very complex, and the combination of the two may produce unintended consequences that might result in an accident or incident.
We have not yet seen any accidents involving either of these aspects of cyber security, but stay tuned.
In conclusion, as the expansion of IT capabilities continues to accelerate, the NTSB continues to keep abreast of these changes and make use of them for enhanced accident investigations. Our training, recognition of previously seen patterns, and collaboration with outside experts, irrespective of whether they are parties to our investigations, will help us ensure our continued ability to investigate accidents involving advanced technologies in the future.
Thank you again for inviting me, and I would be happy to take any questions.