Northeast Regional 188 - Accident In Philadelphia

Greg Moore wrote: Finally, realistically or not, passengers are going to want a real life meat brain in the cab.

Boy, have I worked with a few of THOSE over the years!

Arborwayfan wrote:I'm not quite sure how or when is the time to do this, but at some point some people with credentials to be believed and no connection to Amtrak need to point out to the media and the public in general that (a) train travel is safer than driving even with this accident factored in (b) Amtrak is a safe transportation company (c) where some improvements in safety can still be made and (d) private entities--airlines, buses, etc.--have accidents, too. It's hard to think exactly how to do this without some people thinking it's disrespectful to the dead and injured and some people thinking it's some kind of political ploy, but consider: if this wreck leads to the abolition of Amtrak on spurious grounds, or leads people to drive more, that would lead to more highway deaths even if Amtrak doesn't get safer. I guess that didn't happen after the MN wreck or the one in CA a few years ago, but every time there is a wreck--and fortunately they are pretty uncommon--I worry that because the train wreck is spectacular people will just forget all the car crashes and end train service because they mistakenly think it's more dangerous than other ways of getting around.

Thoughts?

Train travel is so much safer than car travel it isn't even funny. Ironically, I think a big reason for this is auto collisions (no, I'm not calling them accidents because most are caused by human error) are so common place that they're now just like background noise. You drown it out without thinking about it. Arguably, given the number of annual fatalities (~35K in the US alone) if the NTSB were in charge of road travel it would be shut down altogether until a way could be found to make it much safer. We don't need to preach how safe train travel is. We just need to make people more aware of how dangerous road travel is, even to innocent bystanders who are not in a car. Self-driving cars may change that somewhat in the future, but road travel can never come close to matching the safety records of start-of-the-art high-speed railways. Unfortunately, people have the illusion that being in control makes them safe. In reality it doesn't work that way. 90% of the American public lacks the coordination, intelligence, spatial perception, or proper attitude to safely drive a motor vehicle regardless of the amount of training they receive. It shows in the safety statistics.

Maybe we need to have every fatal car wreck get the same sensational coverage a train wreck does. The only problem with that is you would need a few news channels entirely devoted to this given that about 100 people a day lose their lives on the highways. In fact, that's a great way to put this into perspective. More people will die on the highways in 2 or 3 hours than died in this incident. That isn't even getting into the indirect deaths caused by motor vehicle exhaust. By some accounts those are ten times more than the direct deaths.

Greg Moore wrote:

BR&P wrote: Computers generally are far better at routine tasks. Things like the DC metro used to run (and it's returning to) Automatic Train Operations. Yet it still had a driver.
In theory, modern aircraft can back from a gate, taxi, take off, land and find a gate w/o human input. But it's not done.

The problem is the non-routine. Take the example of the Miracle on the Hudson. There were a number of decisions that had to be made that are hard to predict before hand.
Something like a mine line which has basically two endpoints in a remote area is fairly easy to automate. You can predict most surprises and in many cases your simple solution to any unknown would be to dump air, stop the train and wait for someone to drive out there to see what's going on.

In something like the NEC, there's just far too many variables to really automate the whole thing. And, as the Metro crash shows, automatic systems can have tragic failure modes.
Finally, realistically or not, passengers are going to want a real life meat brain in the cab.

One thing you're overlooking is AI is increasing in power exponentially. The problem here isn't that we don't know how to program machines to learn to do non-routine tasks. Rather, it's that we don't have the computing power (yet) for them to learn on that level, at least not in a compact size which might fit in a vehicle. That's coming though a lot sooner than people think. Probably in a decade, machines will be able to learn and do routine and semi-creative tasks about as well as a human child. Within a generation, they'll probably exceed an average human adult. It may be a while before we have AI physicians or engineers doing a better job than humans but mostly rout tasks like driving a train, a motor vehicle, even a plane, will be done within 10 to 20 years as well or better than they can be done by a human who is expert in those areas. At that point, it will be largely a societal decision on whether or not to keep a human body in these jobs basically to just monitor the machine. I could make a great case against it. Why? Humans are horrible at repetitive tasks. They're even worse at tasks they're largely not involved in except in exceptional circumstances. Think of a job like a security guard. You have hours of tedium with literally nothing to do, punctuated by very rare times you need to take action. It would be much the same if we automated our vehicles but kept a driver on board just in case the automation failed. Chances are great the person would either be asleep, or not react fast enough to make a difference. Nothing would be gained in terms of safety. That being the case, you might as well just have 100% automation.

As for any arguments that AI and machines may make people so bored they overthrow governments and the like, I doubt it. The problem is largely one of education. Our education system basically trains people now to be cogs in a wheel instead of training them for creative pursuits. If that were done instead then people will find plenty to occupy their time if they didn't need to do rout tasks 40 hours a week. I think a society where machines do all the grunt work would be an infinitely better one than what we have.

I would thimk the role of technology is to assist humans to operate machinery more safely. I think the most significant safety advance lately, has been the increase and avalaibilty of wifi, and digital cameras. In the case of an accident like this, wether caused by human or machine, the technology is there to alert both the conductors, and the control center, that something is wrong. And to enable both to then intervene, and either stop the train, or take other action as necessary.

David Benton wrote:I would thimk the role of technology is to assist humans to operate machinery more safely. I think the most significant safety advance lately, has been the increase and avalaibilty of wifi, and digital cameras. In the case of an accident like this, wether caused by human or machine, the technology is there to alert both the conductors, and the control center, that something is wrong. And to enable both to then intervene, and either stop the train, or take other action as necessary.

In the short and medium term that's true. We already have collision avoidance in some automobiles. Until AI increases in power we'll still need humans to run many types of machinery but we can have interventions which detect and stop dangerous inputs to the controls. We also can now inexpensively store massive amounts of video to allow us to determine the cause if something does go wrong. One thing to remember though is many tasks which were at one time under human control are now completely automated and nobody even thinks about it. For example, elevators have largely been automated for decades even though at first there was apprehension to doing so. I think the time will come when our children or grandchildren look back and say with astonishment people actually used to drive trains, planes, and cars.

jtr1962 wrote:Maybe we need to have every fatal car wreck get the same sensational coverage a train wreck does.

Sure. Mathematically speaking trains are safer, even though road travel accounts for 4.3 trillion passenger miles per year and Amtrak only 6.8 billion per year. (Commuter rail 11.121 billion, Heavy Rail transit 17.516 billion, Light Rail transit 2.316 billion passenger miles. 2012 figures.) Passenger railroads could kill 295 people per year and still be safer than road travel.

BUT - When we choose public transportation we place our lives in the hands of the operator. Which makes it less OK to kill 295 people per year. Fortunately rail operations only kill around 29 passengers per year (2000-2009 average). A tenth of the highway death rate.

With numbers like that perhaps an actuary would say that further safety improvements are not needed. But as I posted in another thread when the Amtrak 188 accident was breaking news ... raw numbers is not the way we make decisions. We are humans with emotions and we make "irrational" decisions.

29 people per year at the hands of trusted rail transportation is too many. We can work on lowering the number of road accidents as well (especially grade crossing accidents) but I don't consider 29 to be an acceptable number.

David Benton wrote:I would thimk the role of technology is to assist humans to operate machinery more safely. I think the most significant safety advance lately, has been the increase and avalaibilty of wifi, and digital cameras. In the case of an accident like this, wether caused by human or machine, the technology is there to alert both the conductors, and the control center, that something is wrong. And to enable both to then intervene, and either stop the train, or take other action as necessary.

When I see the advertisements for the "smart cars" that automatically avoid accidents with alerts and braking I see it as compensation for the distractions that technology has brought into the car. We started with radios to fiddle with and cupholders ... low tech but still enough technology to get people in trouble as they paid attention to the radio and food and less attention to the driving. We have added even more distractions over the years.

Trains are very controlled environments operated by professionals. There are plenty of rules in place to guide these professionals ... why not give them as much technological support as possible? I do not want to see one man trains or full automation ,,, but support the professional engineer as much as possible.

David Benton wrote:I would thimk the role of technology is to assist humans to operate machinery more safely. I think the most significant safety advance lately, has been the increase and avalaibilty of wifi, and digital cameras. In the case of an accident like this, whether caused by human or machine, the technology is there to alert both the conductors, and the control center, that something is wrong. And to enable both to then intervene, and either stop the train, or take other action as necessary.

The Americans want to run fast trains, but are not willing to spend money building a suitable track. European and Japanese fast trains mostly run on specially built tracks. The east coast mainline is not suitable for fast trains. It has too many curves, junctions and other causes to change speed. In this accident the train was going too fast for the track it was on. The driver needed to slow the train but seems not to have done so. Possibly he was hit by a vandal's missile.

I'm curious to know but does anyone have information as to the specific location where the SEPTA train came to a stop after being struck with debris? I was curious to know more or less where it might have occurred and where Train 188 was at the time this occurred? I remembered hearing reports originally that the SEPTA train was much further south from 188 when it reported having been hit by a rock or shots but wondered at what point would have both trains crossed paths? Obviously the Acela that had its window shattered was noticed by a passenger and probably didn't notify a crew member in a more timely manner to allow one to pin point where the impact occurred, but I wanted to see if it was possible for the source of the debris that struck the SEPTA train to also strike train 188 as it passed by, from say the same overpass? After all, both trains were traveling at speed in opposite directions, so the distance between them grew in a short time frame.

Having a malfunction with your cab signal is not this overly dramatic event, and takes place a little more commonly than one thinks, though it is by no means an everyday event. It would most certainly require much more attention from the engineer, and in the case of the LIRR, the Conductor should be present with the Engineer to call out the signals. It will make your day a little harder, but it shouldn't razzle you to the point where you become disoriented or stressed for the rest of your work day. I'm not sure how it goes on the NEC but I would assume that Interlockings are not closely bunched up in short distances as they are say on the LIRR or MNR. With a cab signal indicator malfunction, we would be required to approach every interlocking prepared to stop. Obviously the signal indication on the signals will dictate your movement, but it will not be unusual to the engineer operating the train because after all, they are always looking at signal aspects regardless of whether they have cab signal indications or not. So I don't think that his southbound trip would have affected him in such a way that he would be distracted on 188.

As the story continues, it still points more and more to a distraction of some sort or a situation where he couldn't control the train in the manner that he should have been. I also wonder how the FBI will be able to determine if the damage to the Fireman's window were made during the crash or before the crash? If it was say a rock that struck it, how will they know when that rock struck the window? I see the damages to the front end that many of you are speaking of, and yes, some of it is related to the crash, But if the camera was reported to have stopped working when the train derailed, then we can assume that it did not film the locomotive barreling down the embankment with objects striking the front end. And since that impact on this specific window was above the area where the camera is, then other than audible recording, how will it show that the glass was struck above it? I'm sure also that the NTSB will be looking into the possibly of what could be a chain reaction event. The train being struck with an object, with a report on the radio of another train being struck, the reporting that 188 was struck too, suddenly 188 was at over 100 MPH entering a 50 MPH curve... A lot of variables here to contend with. The possibility exists that the combo of being hit and the loss of situational awareness all contributed to the event.

justalurker66 wrote:
. . .
Sure. Mathematically speaking trains are safer, even though road travel accounts for 4.3 trillion passenger miles per year and Amtrak only 6.8 billion per year . . .

Fortunately rail operations only kill around 29 passengers per year (2000-2009 average). A tenth of the highway death rate.

Frankly, I'm surprised that the death rates are as close as that; I would have thought that it was relatively much more dangerous to drive than that.

Regarding automation, I think that the computers are way powerful enough to safely run trains already, and that programmers are also capable enough to manage the task (unfortunately, those annoying bugs do pop up and with lives at stake that would be a serious factor). Nevertheless, I think that fail-safe programs are reasonably achievable.

The technical problems that would be holding this back involve data acquisition - the plethora of sensors of various types and the data transmission between the sensors and the on-board computer and central control. Reliability and security in this area does not currently appear to be up to snuff. GPS, pattern recognition, and even radar and laser imaging are all subject to some limitations and/or can be spoofed to some extent or otherwise subverted by hackers or other high-tech equivalents of rock throwing vandals.

Of course, I would be much simpler and arguably better to have a qualified engineer either supervising or being backed up by the AI.

In the future, perhaps the lead car of the train will be a first class view lounge and the engineer, in a role more like the captain of a cruise ship, will be on hand in case of emergency, but mainly to hob-nob with the hoi polloi.

I've thought about this wreck for a week and while all these conspiracy theories about rocks, bullets or other projectiles hitting the loco windshield make for interesting reading simple fact remains nothing short of a 105 round is going to derail an 800 ton train.
For whatever reason, hopefully to be determined by investigation, engineer became disoriented, lost track of his location and sped through a restricted curve, nothing else fits.

http://www.washingtonpost.com/blogs/won ... to-travel/" onclick="window.open(this.href);return false;
Northwestern University professor Ian Savage has the definitive multi-year data on this.

Anything "policy makers" do to make the train more expensive (or "safer") runs the risk of driving people off the train and into cars, where they're 15 to 20 times more likely to be killed per mile.

Frankly, that bus is so low per passenger-mile is a good sign: it says that professional drivers make a huge difference (based on how they outperform cars), and that whatever the problem is with trains, it probably isn't train engineers, but somebody or or some thing trackside.

That isn't to say that we shouldn't be spending Federal "safety money" to get Train down to the Subway/Bus/Plane level (we should, and can). That the Shinkansen have never had a fatality and the TGV have had one say that its possible to make trains safer than air.

mark777 wrote:I'm curious to know but does anyone have information as to the specific location where the SEPTA train came to a stop after being struck with debris? I was curious to know more or less where it might have occurred and where Train 188 was at the time this occurred? I remembered hearing reports originally that the SEPTA train was much further south from 188 when it reported having been hit by a rock or shots but wondered at what point would have both trains crossed paths? Obviously the Acela that had its window shattered was noticed by a passenger and probably didn't notify a crew member in a more timely manner to allow one to pin point where the impact occurred, but I wanted to see if it was possible for the source of the debris that struck the SEPTA train to also strike train 188 as it passed by, from say the same overpass? After all, both trains were traveling at speed in opposite directions, so the distance between them grew in a short time frame.

Having a malfunction with your cab signal is not this overly dramatic event, and takes place a little more commonly than one thinks, though it is by no means an everyday event. It would most certainly require much more attention from the engineer, and in the case of the LIRR, the Conductor should be present with the Engineer to call out the signals. It will make your day a little harder, but it shouldn't razzle you to the point where you become disoriented or stressed for the rest of your work day. I'm not sure how it goes on the NEC but I would assume that Interlockings are not closely bunched up in short distances as they are say on the LIRR or MNR. With a cab signal indicator malfunction, we would be required to approach every interlocking prepared to stop. Obviously the signal indication on the signals will dictate your movement, but it will not be unusual to the engineer operating the train because after all, they are always looking at signal aspects regardless of whether they have cab signal indications or not. So I don't think that his southbound trip would have affected him in such a way that he would be distracted on 188.

As the story continues, it still points more and more to a distraction of some sort or a situation where he couldn't control the train in the manner that he should have been. I also wonder how the FBI will be able to determine if the damage to the Fireman's window were made during the crash or before the crash? If it was say a rock that struck it, how will they know when that rock struck the window? I see the damages to the front end that many of you are speaking of, and yes, some of it is related to the crash, But if the camera was reported to have stopped working when the train derailed, then we can assume that it did not film the locomotive barreling down the embankment with objects striking the front end. And since that impact on this specific window was above the area where the camera is, then other than audible recording, how will it show that the glass was struck above it? I'm sure also that the NTSB will be looking into the possibly of what could be a chain reaction event. The train being struck with an object, with a report on the radio of another train being struck, the reporting that 188 was struck too, suddenly 188 was at over 100 MPH entering a 50 MPH curve... A lot of variables here to contend with. The possibility exists that the combo of being hit and the loss of situational awareness all contributed to the event.

The Septa train terminated at the nearby North Phila station. There also was a window shot at on a car on the Acela train too. There are pics online of both windows.

As I recall, Shinkansen had 1 fatality a few years ago, dragging a passenger to his death when he became caught in a door. Still an incredible record for a system, though.

bluedash2 wrote:The Septa train terminated at the nearby North Phila station. There also was a window shot at on a car on the Acela train too. There are pics online of both windows.

The Acela was held at 30th St. for Amtrak Police to interview the witnesses and the mechanical dept. to come out and examine the window. Arrived in D.C. ~30 mins. late after they all finished up and the window was deemed structurally safe to proceed. It would've been a different story if it were one of the power car's windows, but this projectile hit a passenger window unlike the SEPTA and Regional trains and did not breach the innermost layer of glass.