Skip to main content

tv   NTSB Chair Christopher Hart Discusses Self- Driving Cars  CSPAN  September 6, 2016 11:08pm-12:10am EDT

11:08 pm
me, donald doesn't know anything about the casino big. >> 8:00 eastern pacific on c-span's "q&a." now, national transportation safety board chair christopher hart, he spoke at the national press club on the safety of self-driving cars. this is about an hour. national transportation safety board chairman christopher hart has been in this role for a little over a year, and already has made his mark in a job that can involve chasing the latest problems and being frustrated when it comes to preventing them. the ntsb, with a staff of 400, is an independent federal agency that investigates significant accidents involving railroads, highways, u.s. waters, and airplanes. just this week, the board was called into action to probe a fiery head-on crash between two freight trains in texas. the agency determines the probable cause of the incidents and issues safety recommendations aimed at
11:09 pm
preventing future accidents. in addition, the ntsb studies transportation safety and coordinates the resources of the federal government and other organizations to provide assistance to the victims and their families affected by major transportation disasters. hart is the great-nephew of james herman banning, the first african-american to receive a u.s. pilots license which he got in 1926. banning was killed in a crash during an air show before hart was born but aviation must have been passed through down the family genes. hart is a licensed commercial pilot. hart's career has taken him to the federal aviation administration and to an earlier term on the ntsb. he also worked for the national highway traffic safety administration which may have sparked interest? what he's here today to talk about. autonomous vehicles. it's a wild, wild west of a fill that holds potential for improving highway safety but also poses a lot of questions.
11:10 pm
who is liable if a crash occurs, what happens if a technology that's supposed to help people drive safely have the opposite effect? autonomous vehicles c s can bens from other. most other modes use some form of automation. please help me give a warm national press club welcome to national transportation board chairman, christopher hart. >> thank you, tom, for that very kind introduction. thanks to all of you for coming and certainly thank you to the national press club for inviting me to speak on behalf of the ntsb. it's a privilege and honor to be here. when they invited me, i warned them they were going to have trouble shutting me up because i love talking about this stuff. i'm an attorney. my credo is never use one word
11:11 pm
when two will suffice. driverless cars have been called all kinds of names. i'm going to call them driverless cars. i want to talk about how the ntsb can help the process of bringing them onto our streets and highways. by doing this, i don't mean to suggest that we're looking for work because our plate is already very full, but i am suggesting that we could be a very valuable resource. so to put my remarks in context, i'll follow up with what tom said. he described it pretty completely. that won't stop me from using two words when one will suffice. ntsb is an independent federal agency and oversee accidents in all modes of transportation. a lot of people think of us as aviation accident investigators only but we do all modes of transportation. we do that to determine what caused the accidents then make recommendations to try to revent them from happening again. our primary product is recommendations. our world-class investigators and analysts don't like to give up until they have the answer,
11:12 pm
until they found out what caused the accident. and the recommendations that they create are so compelling that the recipients respond favorably to our recommendations more than 80% of the time even though they don't have to, they're not required to do that. we are not a regulator. we're a recommender. they do that more than 80% of the time. we like to think that the implementation of our recommendations has helped to make transportation safer for all of us. speaking personally, it's a privilege and an honor for me to be here because we do have such world-class investigators and analysts and they do all the hard work and i get all the credit, so what's not to like? my remarks come from the context of our experience as accident investigators. driverless cars are coming. there's no doubt about it. and their potential for improvement is absolutely amazing. first and fore momost, driverle cars could save many, if not most of the 32,000 lives lost every year on our streets and highways. a tragic and unacceptable number
11:13 pm
that's been decreasing for several years but recently as most of you know taken a turn in the wrong direction. driverless cars could also increase the amount of traffic that our roads can safely carry because instead of maintaining a car length separation for every ten miles per hour as i'm sure we all do, driverless cars could reduce that separation. stay tuned for what other amazing changes might be possible. how might that happen? ideally with automation. most crashes on our roads are due to driver error. the theory of driverless cars is that if there's no driver, there will be no driver error. ideally, removing the driver would address at least four issues that are on the ntsb's most-wanted list of transportation safety improvements. namely, fatigue, distractions, impairment, and fitness for duty. the automation in driverless cars would presumably also address another item on our list, improve collision avoidance technologies. decades of experience in a variety of contexts has
11:14 pm
demonstrated that automation can improve safety, reliability, productivity, and efficiency, bull but that experience has also demonstrated there can be a downside. as noted by professor james reisen, record we nounrenowned quote, in their efforts to compensate for the unreliability of human performance, the designers of automated control systems have unwittingly created opportunities for new error types that can be even more serious than those they were seeking to avoid. end of quote. our investigation experience provides three lessons learned that support professor reisen's statement. the first is that the theory of removing human error by removing the human assumes that the automation is working as designed. so the question, as always, what if the automation quits or fails? will it fail in a way that is safe? if it cannot be guaranteed to fail in a way that's safe, will the operator be aware of the failure in a timely manner and
11:15 pm
will be the operator then be able to take over to avoid a crash? an example of the automation failing without the operator's knowledge occurred right here in washington and you may remember the metro crash near the ft. totten station in 2009 that tragically killed the train operator and eight passengers. in that accident, a train temporarily became electronically invisible. we found that there was a -- it was called a parasitic oscillation in the electronics. i minored in electrical engineering and never heard of parasitic oscillation. that's what caused this train to become electronically invisible. when that happened, the symbol of the train disappeared from the display board in the central dispatch center. when a train becomes invisible on the board, an alarm sounds. problem is that the alarm sounded several hundred times a day. so that meant it was largely ignored. unfortunately, when the train became electronically invisible, there was no alarm on the train behind it regarding the electronic disappearance of the
11:16 pm
preceding train. that's why the operator of the train behind was unaware of this disappearance of this electronic disappearance. instead based on the electronically unoccupied track ahead, the automation in the train behind began accelerating to the maximum speed for that area and by the time the operator saw the stopped train and applied the emergency brake after coming around the curve which is what limited her sight distance, it was too late. another lesson learned in support of professor reisen's statement, efsh if the operator is removed from the loop, humans are still involved in designing the vehicles, manufacturing the vehicles, maintaining the vehicles and they're involved in the same functions with respect to the streets and highways. each of these points of human engagement presents yet another opportunity for human error. moreover, human error in these steps is likely to be more systemic in its effect which means it possible involves several vehicles instead of just one and more difficult to find and correct. an example of this lesson learned is the collision of an automated driverless people mover into a stopped people
11:17 pm
mover at the miami international airport in 2008. that collision was caused largely by improper maintenance so even though they had no operator, it still had a crash caused by some other point of intervention of human error. the most fundamental lesson learned from our accident investigation experience in support of professor reisen's statement is that introducing automation into complex human centric systems can be very, very challenging. most of the systems we have investigated, excuse me, are becoming increasingly automated but not yet fully automated. as a result, wefr seen the challenge have been even more difficult in a system that still has substantial human operator involvement and is not yet completely automated. situations involving partial automation with substantial human operator involvement have demonstrated two extremes. excuse me. on one hand, the human is the most unreliable part of the system. so that's the reason for trying to take the human out of the system. on the other hand, if the system
11:18 pm
encounters unanticipated circumstances, a highly trained and proficient human operator can save the day by being the most adaptive part of the system. an example of the human operator saving the day is captain sollenburger's amazing landing in the hudson river when his airplane suddenly became a glider because both of its engines were taken out by birds. in stark contrast, a textbook example of the complexities of the human automation interface in which the human was the most vulnerable part of the system is air france flight 447 from rio de janeiro to paris in 2009. after air france 447 reached its cruise altitude of 37,000 feet at night over the atlantic and began approaching distance thunderstorms, the captain left the cockpit for a scheduled rest break. in doing so, he gave control to two less experienced pilots. the airplane had tubes that project from the fuselage to provide information about how fast it's going. air speed information is so
11:19 pm
important that there were three pido tubes for redone dansy andy were heated to ensure they were not disabled by ice. with abundant supercooled water from the nearby thunderstorms, the po2s were overwhelmed and became clogged with ice so the airplane no longer knew how fast it was going. the loss of air speed information caused systems to quit which they're designed to do when they don't have reliable information. included the automatic pilot flying the airplane and automatic throttle maintaining the speed. as a result, the pilot suddenly had to fly the airplane manually. the laws of air speed information also rendered inoperative the automatic protections that prevented the airplane from entering an aerodynamic stall in which the wings no longer produced list. the pilots responded inappropriately to the loss of these systems and the result was a crash that tragically killed
11:20 pm
all 228 onboard. as with most accidents that we investigate, several factors played a role. to begin with, the redundancy of having three pido tubes was not effective because all three were taken out by a common cause. in addition, the pilots had not experienced this type of failure before, even in training, where the problem can be simulated in very realistic simulators. so as a result of never having seen it before, they were unable to figure out just what went wrong. finally, use of the automatic pilot is mandatory at cruise altitudes so the pilots had never flown manually at that altitude before even in training in the simulator. this is important because the airplane behaves very differently at cruise than low altitudes such as during takeoff and landing. operational and design issues compounded the problem and led to the tragic outcome of the loss of 228 people. as an aside, the pido tubes have frozen before in that type of airplane but the pilots in those
11:21 pm
previous encounters responded successfully. consequently, the entire fleet including the accident airplane was scheduled for the installation of more robust heaters but given that the previous encounters were successful, an immediate emergency replacement was not considered to be necessary. with that background on how automation can be both the good news and the bad news, let me turn to how the ntsb can help inform the process of moving toward driverless cars. first, as i have just explained, we offer considerable experience regarding the introduction of automation into complex human centric systems. most of our investigations involve relatively structured systems with highly trained professional operators who have various requirements regarding proficiency, fatigue, impairment, distraction, and fitness for duty. given that human drivers will probably be in the loop for some time to come, i would suggest that as difficult as that transition to more automation has been in the structured and regulated environment that we have investigated, it may be
11:22 pm
even more challenging in a public arena in which drivers are usually not highly trained and may be fatigued or impaired or distracted or not medically fit. whether some human drivers will always be in the loop because they would rather not use the automation for various reasons, for example, they just don't trust it or they just like to drive. the second way that the ntsb can help relates to collaboration. the auto industry has already recognized the importance of collaboration as most recently shown by the collaborative approach regarding autonomous emergency braking. our experience with collaboration, especially regarding commercial aviation, may help improve it further. so let's talk about where we've seen collaboration in aviation. the most recent fatal u.s. commercial airline crash occurred in 2009, and more than once in recent years a commercial aviation industry has gone years in a row without a single passenger fatality. although automation has played an important role in the industry's continuing safety improvement, much of the industry's exemplary safety record is attributable to
11:23 pm
collaboration. in the early 1990s after the industry's accident rate had been declining rapidly, the accident rate began to flatten on a plateau. meanwhile the federal aviation administration was predicting the volume of flying would double in 15 to 20 years. the industry became very concerned that if the volume doubled, while the accident rate remained the same, the public would see twice as many airplane crashes on the news. and at that point, it doesn't help to go to the public and say, don't worry, the rate's real low. what the public counts is the number of times they see crashed airplanes on news. that caused the industry to do something that's never been done in any other industry before or since, pursued a voluntary collaborative industrywide approach to improving safety. this occurred largely because they realized the way to get off the plateau was not more regulations or bigger shtick for the regulator, but instead the way was to figure out a better
11:24 pm
way to improve safety in a very complex aviation system. the voluntary collaborative process known as c.a.s.t., commercial aviation safety team, brings all the players, the airlines, the manufacturers, the pilot, the air traffic controllers, and the regulator all to the table to do four things. first, identify potential safety issues. second, prioritize those issues because they quickly realized they'd be identifying more issues than they would have resources to fix. third, develop interventions for the prioritized issues. and fourth, evaluate whether the interventions were working. the c.a.s.t. process has been an amazing success. it resulted in reduction of the aviation fatality rate from that plateau on which it was stuck, reduction from that plateau by more than 80% in less than 10 years. this occurred despite the fact that the plateau was already considered to be exemplary and many thought the rate could not decline much further. the process also improved not only safety but productivity which flew in the face of conventional wisdom, that
11:25 pm
improving safety usually decreases productivity and vice versa. in addition, a major challenge of making improvements in complex systems is the possibility of unintended consequences. yet, this process generated very few unintended consequences. and last but not least, the success occurred largely without generating any new regulations. as an observer in c.a.s.t., the ntsb can help determine how much this aviation success story is transferable to their industry. one size may not fit all. the airlines do not compete regarding safety. you've never seen an airline ad saying we're the safest out there. you always see ads by auto manufacturers that our car's the safest. so the 80% reduction in the fatality rate accomplished by c.a.s.t. even though one size may not fit all is a very powerful example of how much can be accomplished relatively quickly through voluntary collaboration. another difference between the
11:26 pm
two industries is that the aviation regulatory framework is largely federal, whereas collaboration regarding driverless cars would probably need to include significant participation by the states. the third way that the ntsb can inform the process of introducing automation relates to onboard event recorders. our investigations -- excuse me, our investigation are significantly enhanced when we have event recorders to tell us what happened. airliners have had black boxes which, of course, are actually orange. they've had black boxes for decades to record both the aircraft parameters and the sounds in the cockpit. other transportation modes are increasingly introducing event recorders as well as audio and video recorders. assuming that difficulties will be encountered as automation is being introduced, the more the industry knows from the event recorders about what went right and what went wrong, the more the industry will be able to fashion remedies that effectively address the problems. accordingly, consistent with another item on our most-wanted list, expand the use of recorders to enhance
11:27 pm
transportation safety. we would encourage the use of robust onboard recorders to help the process. event recorders and oa erers in of transportation introduce significant issues, privacy and the appropriate use of recorder data. the ntsb's sensitivity to these issues already helped to inform the conversation in commercial trucking and the process of improving passenger vehicle event recorder as wheell pel. in closing, rathser than waiting for accidents to happen with driverless cars the ntsb engaged to help inform how driverless cars can be safely introduced into america's transportation system. our experience in the introduction of automation into human-centric systems, our appreciation of the power of collaboration and our understanding of the importance of onboard event recorderses all position the ntsb to provide valuable assistance in this process. so thank you again for inviting me to speak today. i would be happy if i have time, and i think i do, to answer any
11:28 pm
questions. thank you, tom. >> thank you, mr. chairman. we do have a lot of questions for you today, so i hope you're ready. >> good, this is my favorite part. >> good. let's start off with a pretty easy question e. what scares you the most about autonomous cars? >> that's a good question, but there isn't any single thing that scares me the most because the whole process is going to be very complicated. i mean, there are -- i think people are wildly underestimating the complexity of bringing automation into this system. so there's not one things it's just sort of the total picture. is unnerving to me and i think we have a great opportunity to help because we can transfer the wonderful success story from other modes, mostly aviation to help it happen better in this body. >> u.s. automakers have huge hopes autonomous cars will improve safety, is there a worry that the first fatal crash involving a self-driving car may
11:29 pm
bring the whole enterprise down? isn't it likely that people will overreact to this, despite the fact there are 30,000-plus highway fatalities that they're used to every year? >> i think it's fair to say it will certainly -- the first fatal crash will certainly get a lot of attention, but this train has left the stationing. i don't think it's going to be stopped by one or even two, if there's a trend, that's another thing, but just there will be -- there will be fatal crashes, that's for sure, but i don't think it's going to be stopped just by a crash here or there, and especially because it's probably still happening at a lower rate than what it happens without automation. >> what's the best way to demonstrate autonomous vehicle technology as safe? enough to be allowed on u.s. highways? should google cars be held to a higher standard, for example, than human-driven cars? >> well, that's an interesting question. i go back to the collaboration that's going to be important. so many people are going to have to work together. the moral of collaboration is very simple. anybody who's involved in a problem should be involved in developing the solution. this is going to take collaboration of a lot of people
11:30 pm
to make sure that what they're doing enhances safety and generates safety as much as possible. >> on that note, driverless cars are likely to be on the road before states have regulatory regimes in place to govern them. what are the first steps states should take to handle regulatory complexities especially with the mix of human drivers and driverless cars on the road? >> well, i'm going to return to this topic many times, i'm sure, but it goes back to collaboration again because we -- it will not end up as a patchwork quilt of this state you have to have your hand on the wheel, that state you don't have to. there's going to have to be some uniformity across state lines and that -- and collaboration is going to help generate that. so i return to the issue of collaboration. it's going to take a lot of people working together and this will be much more complicated than in aviation because aviation, it was federal, so we only had one source of legislation to look to, whereas in states, it's going to be significantly more complicated enterprise. >> let me follow up on that. already in the united states, we
11:31 pm
see a patchwork of a lot of laws governing different things. states sometimes don't work together. so is there -- do you fear the idea of driving from utah to colorado or something like that where there's a difference and all of a sudden you're breaking the law when you weren't in another state because they didn't work together? >> that's a fair question. i think it's fwoings going to b evolutionary process that goes back to, guess what, collaboration. it will be a collaborative process where the states are realizing how much it's going to hurt their commerce if they don't join this effort collaboratively. >> we're getting handed, like, a million more questions. this is fun. >> you're not having as much fun as i am. >> i'm interested if learning more about what steps the ntsb has taken regarding this relationship between self-driving cars and moetser cycles. specifically is it safe to assume that self-driving cars can detect motorcyclists and if so, what specific precautions or scenarios are used to account for individual riders on motorcycles as well as groups of
11:32 pm
motorcycles? >> i can say as a matter of my experience in a very general way that the driverless cars are made to recognize anything out there that could be an issue. so that could be a pedestrian, it could be a deer, it could be anything that moves that is a potential problem and motorcycles would be part of that as well, so they wail have to -- the driverless cars will have to figure out how to operate in the environment they are in and that's one of the big challenges is because that environment is so variable. >> do you expect some pushback from, let's say, municipal governments, that count on speeding tickets for revenue? >> good question. next question? >> we'll put that down as a no comment. what is your prediction on the rate of adoption for driverless cars? >> i think that people are, as i said before, underestimating the complication of making this happen. especially if you talk about when are we going to be totally driverless except for the few
11:33 pm
people like me who just like to drive? so i think that's going to take longer than people are thinking and i think it's going to be a far more complex effort than people are thinking, so it's hard to put a time on it. i can just say that if we inform the process with the success stories from other modes like aviation, it's going to be a much smoother ride. >> we were chatting before the lunch, i was told by this time at my age, i would have a flying skateboard, at least, right? and flying cars. how do you educate the populous about this is not necessarily science fiction, it's not going to work like it does in "back to the future," other movies like that? >> education is going to be a big issue. i mean, go to the basic fundamental question, does my 13-year-old daughter need to even learn how to drive? or will it be driverless by the time she's driving? probably not in her case because that's in three years but for someone who has a 1-year-old, that's a real question, do they even ever have to learn how to drive? that's a complex question that's going to -- that is going to
11:34 pm
require this collaborative effort in order to address some of these big issues. >> sticking on pop culture for a second, from a nontechnical point of view, how do you see driverless cars changing soci y society, culture, infrastructure? >> i mentioned the infrastructure change because i think the ability to have cars be much closer in spacing is going to hugely increase the efficiency of the use of the infrastructure. so i see some big differences there. i'm wondering, i mean, the social scientists are looking at all kinds of ramifications of this like, for example, do i even need to own a car? or will i just call a car and say, i need to go to work, so a car will show up at my door and i'll take it to work then i call a car, i need to go home from work, and i need to go get my groceries and i need to do this, need to do that. well, that could be a huge change in the way our society works because will individual car ownership be necessary anymore, and will that actually increase efficiency of use of our resources because now
11:35 pm
instead of your car being used an hour a day and parked in a garage for the rest of the time, now the use of the car is going to be much more of the day and so it will be far more efficient utilization of our resources. there are so many potential variations on that theme that i couldn't even begin to know where this exciting concept is going to go. >> you talked about this a little bit, but i want to pin you down if i can. do you think autonomous vehicles should have a licensed driver in the driver's seat or do they have potential as you just mentioned as kind of a carpool drivers for children or chauffeurs or elderly people or others who can't or don't want to drive? >> the answer to that question i think varies with time. i think eventually when this whole thing is figured out, that's going to be some task, you won't need someone who -- i mean, you can be drunk now and you can have a driverless car take you someplace because you're not going to do any of the driving. so, you know, there's already talk about car without the brake and without the steering wheel, so in time we're going to reach a time when many of the cars are that way, but that's not going
11:36 pm
to be any time soon. meanwhile, humans are going to have to play a significant role and that raises the challenge of if your car is mostly driving itself, and then it gives up because you're on a rainy street and the rainy street covers the lane markings so it can't see where the lane markings are, will it tell you you better take over because i can't find the lane markings in time for you to effectively do that? that's one of the big challenges of automation is how does the operator know when they need to take over and will the operator at that point be able to take over? >> you touched on this for a second, but is the bigger benefit of autonomous vehicles the safety on the road or the economy or changes in the savings on infrastructure, for example? >> i can speak to the safety aspect because that's what we do is safety. when i see the possibility of saving 32,000-plus lives a year, a number that as i said is now starting to go the wrong way, that is amazing. so that's what i'll speak to is the safety aspect. that's why we're so interested in this and see an amazing opportunity for us to inform the process with what we've learned
11:37 pm
in other modes of transportation. >> the ntsb of courbviously doe get involved with the vast majority of vehicle wrecks, but as self-driving cars become more common, what will the board's role be, if any, in reviewing wrecks involving automated vehicles? >> what we try to do in highway crashes because as you said, there are so many more than we have staff to look at, we look at the ones that have systemic imply k llications as opposed t had a bad day. that's probably where we'll head with driverless cars as well, we'll look at the accidents that have systemic implications that give us an opportunity to inform the process and make it better. >> and you raised this during your address earlier, one of the contributing factors of the aseana crash landing a couple years ago in san francisco was the pilots became too dependent on automation. how do you guard against people kind of losing the skills they might need to drive with the rise of autonomous vehicles? >> and that's another transitional effect that as we transition from no automation to
11:38 pm
full automation, there's the vast area in between where it's a combination of automation and driver and that's when that issue is going to be important, and it's just going to take some experience and hopefully learning from other modes like what we have seen to make sure that when, if there's a situation that the driver needs to take over, the driver -- the alert is clear, the driver knows that it's time to take over and the driver is in a position to take over. >> so we're talking a lot either autonomous vehicles or self-driving cars. but what about self-driving trucks? and i'm thinking the questioner might be referring to commercial vehicles, i assume. >> i've seen that already that they are talking about self-driving trucks and several, like, amusement parks have se d self-driving buses already. that's a possibility, self-driving -- all the vehicles on the highway are self-driving. i see that has a definite possibility. >> this gets to something you probably won't be able to answer. we're going to try, anyway. do you have any insights on the
11:39 pm
ethics of autonomous car decision making? algorithms to making a decision to save the riders in said vehicle or other cars or pedestrians? >> that's going to be a major question. we'll have to play that one by ear as the issue arises. the example that is given to me oftentimes is you got an 80,000 pound truck coming at you so is your driverless car going to run into the 80,000 pound truck or go on the sidewalk where pedestrians are? a question of you aor me. will the driverless car have a pick me/pick them button? i doubt it. i'm being flip on that. that's a serious issue, today if a driver chose to go up on the sidewalk to avoid an 80,000 pound truck, the driver would probably not face charges for avoiding hitting the truck because that would be clearly fatal for the driver. if the software makes that decision, how is that going to happen? one of the vast array of legal
11:40 pm
and ethical issues that are going to have to be addressed, again, through the collaboration. that's why this collaboration is going to have to include people who are in the law enforcement community. it's going to have to include a whole variety of people that, as i say, if they're involved in the problem, they need to be involved in the solution. >> on that note, as we started off in talking about this, who is responsible in a crash? is it the driver? is it the car? is it pushing the button between choose me or choose them? >> well, i am a lawyer and i'd be happy to go after the legal question, but it's not in our lane, so i'm going to pass on that. >> okay. but on that note, still, will driverless cars require some legislation precluding class-action lawsuits in this case? >> i see a variety of types of legislation that are going to be necessitated by driverless cars. it's going to be a huge shift for everybody, and if i could say, right now we got the difference, some states say you have to have your hands on the wheel, other states don't. so right there is a need for some changes.
11:41 pm
there's going to be a lot of legislative action on this one. at the state level. maybe the federal level as well. but definitely the state level. >> the full lawyer employment act essentially. does the ntsb need any statutory authority to take a role in regulating self-driving cars? >> i'm not able to answer that question well. i don't know what their statutory authority is, what the limits are now, but it would not surprise me if there's going to be some federal electilegislati changes as well. >> as cars become more connected to the internet, do you think there should be -- should ever be a reason that drivers should be checking social media like facebook or sending e-mails or texts? and if not, should the government step in to ban such activities as part of a car's infotainment system or left to the automakers? >> that's a today question. when the car is driverless, that will be a moot point. today, we recommended hands-free cell phone use should be banned in all states. most states ban texting because
11:42 pm
they know what a huge accident cause that is. national statistics say when you're texting, you're 23 times more likely, not 23%, 23 times more likely to have a crash than you're not. almost all states prohibit texting. many states prohibit handheld cell phones. no state prohibits hands-free cell phones. we think even with hands-free cell phones, your mind is on your call and not on your driving task. people say, well, how is that any different than talking to the person next to me? it's very different because the person next to you is another set of eyes who knows you're in a construction queue, on an icy road. the person on the other end of the phone has no idea of your environment. it's hugely different than speaking to the person next to you. >> what kind of data would the ntsb like collected? what about the privacy concerns of consumers with these recording devices? >> privacy issues are ones we've
11:43 pm
had to deal with big-time in aviation. as i said, they've had the black boxes in airplanes for decades. so far, we have not had any breach of information that created a private ssy concern, we've, the industry, has shown an ability to use that information carefully. we use it only for one purpose and that is to figure out what caused the accident so we can try to keep that from happening again. what the owner of the box does, in this karks the owner would be the airline. that's up to them. the law doesn't allow us to do anything with that except use it to improve safety. >> so i'm just laughing about one of the questions i got. we'll ask that one later. you referred to clab rollaborat several times. what kind of collaboration is already happening and is what's happening, what's needed to happen? >> the example that i used for collaboration was regarding the autonomous braking and the agreement to institute it voluntarily. we applaud that because we know from huge experience in aviation that this 80% reduction in the fatal accident rate was prima primarily result of that amazing
11:44 pm
collaboration. we know how powerful vol terrell collaboration can be and i think this agreement, which reaches over 90% of the cars that are being made, that is huge to be able to have it happen as quickly as it did. it would never have happened that quickly if it had awaited a regulatory result. >> moving on to a little different subject. what is the best way that you see to reduce u.s. vehicle crashes? >> automation. no, we -- for about -- just to be more specific, for about 20 years, we've been pushing for something that is a collision-avoidance system, "a," a warning and "b" a way to stop a collision from occurring. we've had recommendation on that subject and subjects like it for more than 20 years. certainly step number one is collision avoidance technology that prevents cars from hitting each other so that's one of the foundation stones for moving to driverless cars.
11:45 pm
>> let's talk about trains for a second. despite several high-profile fiery crashes involving oil trains, such trains continue to run through the heart of certain american cities. such as philadelphia or here in washington, near the united states capital. should trains carrying crude oil be banned from traveling through densely populated downtown areas in the united states? >> after the discovery of oil in north dakota, we all of a sudden saw a lot more crude oil train derailments, and historically before that, if a train derailed, it might be a mile-long train with five or seven or ten cars of crude oil. now it's a mile-long train of nothing but crude oil. if that train derails and one car breach, es, it only takes o car and puts product out in the environment, something ignites that, the other cars even if they don't breach by puncture, are in a thermal environment that encourages breaching. step number one is to keep the trains on the track. we've pushed hard for that. step number two, have more robust tank cars. step number three, address the
11:46 pm
emergency response community because a lot of these accidents happen in the middle of nowhere where it's just a volunteer fire department what hasn't ever seen a hazardous material spills. we started pushing that big-time because of the amazing increase in not only carriage of crude oil which happened because of the north dakota find then also we have an amazing increase in ethanol which became because of the law that said we want to reduce our dependence on foreign oil so we're going to start putting ethanol in the gasoline. we have trains from the corn states to the states that make the oil, so those two things caused us to see a huge rise in trains carrying hazardous material and the cars they were carrying them in were the same cars that were used to karr carry corn oil. that's completely unacceptable. there's a transition toward much more robust train cars. the first step is to keep the train on the track. talks about doing everything you can, positive train control, to
11:47 pm
keep the train on the track. >> speaking of positive train come, there was another train crash this week that may have been prevented by positive train control. are you frustrated railroads have gotten congress to extend the deadline for them to have ptc, positive train control, in the places where passenger rail safety requires them to? >> we've been pushing positive train control since the late 1960s. it's been on the list almost continuously ever since. we took it off briefly when congress enacted a law that said you must do this by the end of 2015. we said, okay, we've taken care of that problem. guess what, they hadn't, because very few people finished it by 2015. yes, it is an ongoing challenge for us. we don't call it frustration. we call it a challenge. that's what keeps us going is we know we have the topportunity t move the needle. >> here in washington, as we all know, the metro has a few issues, but --
11:48 pm
>> myself, today. >> you were early, though. >> go figure. >> but metro has been using manual operations for a while. what is the ntsb's view on using -- >> we're starting to see that. the first place where we saw it big-time was in airport trains. most airports you go to now, the trains are typically operatorless. i'm old enough to remember when elevators had operators. that shows how far back i go, but what we look at is what's the safest way to do it? if it's automation, we're in favor of it. if it's not automation, we're in favor of it. havi to their credit, they stopped using it when it wasn't working properly, so, of course, when you go manual, that means you're going to have more jerky ride. the starts are more jerky, starts are more jerky. i pacompare it with the automat trains i've ridden on. they're very smooth. elevators stop and start smoothly. our concern at the ntsb is
11:49 pm
safe safety. what's the safest way to do it? to their credit, when they realized it wasn't doing as it was supposed to, time to go back to manual. >> stays on metro for a second, how would you praise the performance of the fta thus far in wmta's safe tracking program? >> we have a recommendation after the smoke event in the fontplaza station in january 2015, we put out a recommendation that we don't think the federal transit administration is well suited to oversee this transit property. the reason is because for most of federal transit administration's existence, they had no safety authority what savr. they were basically a funding agency, gave out grants to build infrastructure. so for most of that time, they had no safety authority. after the 2009 ft. totten accident, we said this agency needs safety authority because the transit in the u.s. have no federal safety oversight. we recommended that fta seek that safety authority, and they did, so since most states
11:50 pm
already had state regulatory mechanisms like new york, illinois, pennsylvania, california, texas, florida, they already had state level, so when they created the fta's safety authority, they said let's not dismantle this whole state system because it's working pretty well. let's let the fta work through the states. so that's easy when there's only one state involved. in a few situations, there are two states involved. but in those situations, there's a handful of them, they have reached agreement, i'll take the back seat, you take the front seat. our jurisdiction has three. maryland, d.c., and virginia. and those are three difficult cats to herd. so getting them to work together and getting fta to work through those three states isn't -- hasn't worked so far. we're trying to look for an immediate solution and our view on the immediate solution is to let the federal railroad administration do it because they don't work through the states and they can do it more directly than federal transit administration who has to wait for the three jurisdictions to enact legislation and have an agreement to say that they can work together and that might
11:51 pm
happen by the time my 13-year-old daughter graduates from high school. so we want more immediate action and we think the way to get that and our recommendation says it, let the federal railroad administration oversee this property because this property has to report to three jurisdictions and that is simply, we're the only one in the country that way with three, and that is simply not working. >> the ntsb, under your predecessor, debby herzman, gave a lot of attention to drugged driving as distinct from drunken driving. now that marijuana is legal in many states, does there need to be a legal national limit like there is for alcohol to decide who's too impaired to drive after ingesting marijuana? >> we are very concerned because in every mode of transportation, we've seen a troubling uptick in the use of drugs. my state, colorado, is one of them that legalized marijuana, so we are very concerned that that's going to cause us to see more use of drugs in transportation accidents. i'll give you one very troubling example and that was a truck accident where an 80,000 pound
11:52 pm
truck crossed over the median on the interstate, hit an oncoming bus. we found paraphernalia in the truck. that's the only way we knew this truck driver was using a synthetic drug. guess what, this synthetic drug is one that's available legally over the counter at a truck stop. that's very troubling. something that is an impairing drug is obtainable legally on the highway at a truck stop. so we're very challenged with drugs because there's -- these synthetic drugs, we don't have good understanding of how they work. we don't know the met tab liftic history in the body. we can now look at alcohol and marijuana and tell from a backward look, reverse engineer what was the state at the time of the accident. we don't -- most of the drugs, we don't have that knowledge, we don't know how the drugs interact with each other. we don't know how they interact with alcohol, how they interact with fatigue. there's lots of issues regarding
11:53 pm
drugs that are very troubling and we're concerned and seeing a troubling uptick in every mode of transportation. >> what do you see generally as the board's greatest weaknesses and needs and how do you think that will be impacted by either secretary clinton or donald trump in the white house? >> our challenge is always trying to figure out where -- what's the best next direction to go. so when all of a sudden a lot of oil trains are coming out of north dakota, we're seeing a lot of derailments. our rail staff was not prepared for that sudden spike in our workload. so our challenge is always trying to be strategic and figure out what's the next thing we need to do so we're ready to handle that variability when it arises? >> in your time, since this is a regulatory agency, which recommendation -- agency -- which mode of transportation has been most receptive to following your recommendations? >> that's a good question. number that i mentioned, more than 80% of the time our
11:54 pm
recommendations are responded to favorab favorably. i'm afraid i don't have a breakdown on how that goes from mode to mode. it's probably fairly universal across the modes. i'd have to get back to you with a breakdown on that. >> okay. let's talk about those who don't. why do you think some have not been receptive? >> congress created us to be -- we look only at safety. we do not do cost/benefit studies. so the regulators, they have to look at the total picture and not just the safety picture. they have to look at the totality of circumstances. we're supposed to be the ones who provide the answer in an ideal safety world if safety is your only consideration. i give kudos to congress for creating us that way because what that means is that if the regulators in the industry were doing 100% of what we recommended, that means something is probably broke. that means we're not being
11:55 pm
safety-only enough, or that means the regulator is being safety-only too much. on the other hand, if 60% of our recommendations were being responded to favorably, probably something is broke there, too. i think safety experts would agree around 80%-ish is about right to show the tension that congress intentionally created between us and the industry is working. so, yes, we have to have our hand on the pulse of the economic reality, so, for example, when we first recommended ground proximity warning systems on airplanes, that's a system that warns the pilot, you are approaching the ground too fast under the circumstances, you better do something about it. so when we first started that recommendation, these were big, bulky expensive things, okay, you can put them on a 747, but you're not going to put them on a 19 seat beach 1900 because it's too big and expensive and bulky to put on the little airplanes. when we first made that recommendation, we limited it to the big airplanes. then as the technology improved and they got smaller and cheaper and better, then we started recommending them on all
11:56 pm
airplanes. so no third decimal point cost/benefit study but we had to have our finger on the pulse of economic reality. >> so let's talk about this for just a second. do you think the ntsb should be given more than advisory power? >> that's a good question and people ask me that a lot. they say why not mandate what the ntsb does? i think the reason our product is so good is precisely because it's not mandatory. i think that means our staff knows that if this isn't a real good idea, people are going to ignore it because they can. well, people don't ignore it more than 80% of the time and that's, to me, if we -- if it was mandatory then there's something wrong with telling us we have to be safety only, because if we're safety only and it's mandatory, that's a disconnect. i think it's a good idea that we aren't mandatory and i think that's one of the reasons our product is as world class as it is because our staff knows that if it's not a really good idea, people just won't do it. >> i think you told me your term is up in march but you've been at the ntsb for a while. now that obama's presidency is
11:57 pm
almost done, how would you assess his administration's record on transportation safety? >> in general, all of the industries are improving in their sastty. that's real good news. one of the problems we've seen that's a generic problem across all the modes is regulations have to go through the office of management and budget, and the office of management and budget has a cost/benefit test. so as all the industries get safer, we know that that filter is getting ever more challenging. when we send a recommendation that says require, that's ntsb speak for promulgate a new recommendation, then many times we're told, we're so safe that omb says where are the dead bodies and there may not be any or enough then that stops the recommendation. in a bigger-picture sense, i think we need to have a conversation about, yes, there needs to be some kind of a cost/benefit filter because we don't want money spent willy-nilly without having a clear benefit to it, but i think we need to have a conversation about how to update that
11:58 pm
cost/benefit test to make it more realistic. so, for example, the cost/benefit test cost a split in the fatigue requirements in big airplanes. so if you're flying a big passenger airplane, you have different fatigue requirements than if you're flying a big cargo airplane. our view on that is a big cargo airplane is in the airspace and if it collides with another big airplane, it's going to be just as bad if it's carrying cargo as if it was carrying passengers. but the cost/benefit cost said, well, how many cargo airplanes have we seen that crashed due to fatigue as opposed to how many big airplanes have we seen? i think it's a different way to ask the question and i think the process can improve by being updated with the new realities. >> since you're here at the national press club, i thought i'd ask you, any advice to reporters covering accidents being investigated by the ntsb? >> most of the industries we deal with have a huge problem with the media because of a strong emphasis on sensationalism. an example that's somewhat dated
11:59 pm
now but i'm not sure if it's unrealistic anymore. when airplane crashed, americanairlines crashed in colombia in 1996 and it was a very difficult environment, took a long time for people to get there so when they got there, the bodies had been there for some time. okay. so when they got there, they said they found alcohol in the pilot's blood. well, and so that was big front news -- big front-page news, alcohol in the pilot's blood. four days later on page b6 was, oh, this alcohol was putryification alcohol, not consumption alcohol. meanwhile, the pilot's reputation had been sullied, airline's reputation had been sullied. the public is thinking, oh, this pilot was trunk. the accuracy of the coverage is crucial and we used to be in control of the information a lot more than we are. we're not so much control of it anymore. for example, the one you mentioned in san francisco, the asiana crash, that crash was actually on youtube from the guy who filmed it across the bay
12:00 am
before we were formally notified there had been an accident. so so much for being in control of the information. then to add to that complication, so there was that video. captions copyright national cable satellite corp. 2008 captioning performed by vitac
12:01 am
12:02 am
12:03 am
12:04 am
12:05 am
12:06 am
12:07 am
12:08 am
12:09 am

23 Views

info Stream Only

Uploaded by TV Archive on