911 should be able to stop a train

It was over 5 years ago that I blogged about a robot that would travel in front of a train to spot cars stuck on the tracks in time to stop.

I recently read a local story about an RV that was demolished while stuck on the tracks here. The couple had time to talk to 911, who told them to get out, and it's not clear from the story but it seems like a moderate amount of time may have passed (a couple of minutes) before their RV was smashed.

Here's what should happen, and perhaps it does happen in some places:

  1. The 911 service should receive GPS and cell tower location on the caller. The moment the caller indicates they are stuck on the tracks, the 911 operator should push a button which figures out which tracks it might be and which trains might be approaching that crossing.
  2. Ideally trains are reporting their location with GPS as some do, but schedules can be used, or all trains anywhere near the area can be alerted.
  3. Signal lights close to the crossing should immediately go red, and cell phones of operators on the relevant trains should be called, and the computer or 911 operator can indicate which crossing is blocked. If the engineer is approaching that crossing they can emergency brake.

This can be enhanced a few ways:

  • Each crossing can have a big sign, "If stuck, get out of vehicle immediately, clear track (show direction) and call 911, and give this crossing number NNN." The crossing number would work even if GPS and cell towers don't locate the crossing.
  • Alternately, there could be a 10 digit phone number, different for each crossing. There is, however, some risk of abuse and false reports. You don't want a war dialing telemarketer to stop trains. An operator may still need to confirm.
  • As noted, the sign should try to tell people to clear to the area slightly "upstream" (ie. towards the oncoming train, but not on the tracks, obviously.) That's because when the train hits the car it throws it sideways and forward, never backwards along the path the train came from.
  • If you don't see or hear a train, it makes slight sense to get out and call while walking so the call comes sooner. If you can see the train they can see you and it's probably too late anyway. But human safety is more important.
  • The trains may have another way to reach the engineer, such as a private radio system, but just having a cell phone on each train (plus knowing trains staff personal cell phones and calling all of them) seems like a quick and easy solution. The cell in the train can have a very loud and flashing ringer, especially if it's an emergency call.

It takes a long time to stop a train, but I bet most vehicles that get stuck on the tracks are stuck minutes before the train comes.

Comments

The trains may have another way to reach the engineer

Good old-fashioned text pagers?

They may have pagers but because pagers often take 30 seconds to a minute to deliver pages (or at least have no guarantee of quick delivery) cell phones or radios are probably better.

Note that while one way to do this would be to have the 911 operator call an emergency desk at the railway, this requires the railway to have an emergency desk, and requires an extra step. Using cell phones just means the railway provides a database of train schedules or positions, and phone numbers. Put the phones into any GPS location based service and the railway doesn't have to do anything -- the 911 computer can query where the trains are and phone the trains near the crossing.

More pie-in-the-sky thinking, unblemished by real-world
failure or cost/benefit analysis. Why not just mandate that
all vehicles can no longer fail at railroad crossings, or
place small vials of faery dust at every crossing to be
sprinkled in the case of emergency?

Just placing signs at every crossing -- once the appropriate
regulatory bodies have given their approval -- would cost
millions of dollars for most states. Who do you propose is
going to fund that effort?

And of course the railroad would have to issue cell phones to
its train crews -- you can't rely on the use of personal
phones for this. You're also assuming that cell phones and
GPS never fail and work in all locations. I can imagine many
remote crossings where an approaching train would be out range
of cell towers. (And yes, I know the origin of Sprint.)

And what magic do you suppose is going to take place when the
911 operator "pushes a button that figures out which trains
may be approaching?" Do you expect all the railroads to
continually supply detailed timetables to every E911 agency?

Such crashes may be spectacular, but given the huge number of
vehicle crossings at railroad tracks, they are statistically
insignificant events. It would be far, far cheaper for the
railroads to simply replace the vehicle with a new one of
the victim's choice, no questions asked. More expensive, but
still cheaper than your proposal, would be an advertising
campaign to spread the message that people who get stuck on
railroad tracks should immediately get out of their vehicle
and move away from the tracks diagonally, in the direction of
an oncoming train.

While I agree with you that government planning could indeed make this be much more expensive than it needs to be, it doesn't have to cost that much. A private entity could do it quite cheaply. Indeed, it's all doable with a program running in the 911 operator's computer that is able to map location to train schedules or train GPS data (for the train lines that are doing that and making it public so riders can know if trains are on time) where this exists.

As to how to reach the engineer, well, many trains already have some way to do that, but cell phones seem a good way. Cell phones are very inexpensive, especially compared to the cost of a train and train operation. They don't issue the phone to the crew (unless they want to) they put it on the train.

I am not assuming this never fails. The current system "fails" all the time, there is no warning given to the train crew at all in time for them to stop the train. For commuter trains and urban crossings, they will always be in range I suspect.

As for the benefit, this is not so rare as you say. Quite often people do get out of the car so there is not a death, and the train is solid enough that it usually takes only minimal damage from a car, though the car is totaled. (But cars are totaled all the time.) However, when it happens, the train and the line tend to have to shut down for a long time. I've been on a train that hit a car. You don't even feel it on board the train if you aren't at the front, but you do feel that you sit there for an hour.

The problem is bad enough that many lines are working to spend huge amounts of money to eliminate grade crossings, and feel they must do so if they want to have their trains go faster or more frequently. In the rare cases where new lines are planned for commuter rail, grade crossings seem to be a no-go.

There are better ways to assure this, some of which I'm working on.

One improvement (I invite you to read up on the technology) involves the various flavors of positive train control. To get the required resolution for effective PTC, you need better defined accuracy than current flavors of GPS provide with high confidence. This implies that an appropriate network of HA-NDGPS or equivalent will be put in place... including dead spots for conventional broadcast at the HA frequencies... to cover all the track miles of a system using PTC.

Most PTC systems also require what is called 'civil' input (for example, knowing where maintenance crews or trackworkers are, or where slow orders or other restrictions are in effect). This gives a proper framework for relatively secure input to the train-control system (which among other things is capable of producing orderly stops in minimum distance, not the unmodulated stops of older ATC systems).

There are already many grade crossings with toll-free numbers posted on them -- numbers that go directly either to the railroad police or to operators who know how to address calls. There is NO NEED -- I repeat, NO NEED to involve a perhaps already overloaded 911 operator with a verbose and possibly panicky description of a problem at a crossing the caller probably can't identify, when there is nothing whatever that the police or fire or ambulance crews can do particularly well in the few moments that may apply. Yes, many 911 systems have hot numbers set up to reach railroad police... but it's an additional level of dereferencing for the 911 system, and how does the caller know which railroad they're stuck on (we routinely have people write letters to the editor about trains stuck on crossings that get the railroad name wrong).

Meanwhile, a greater (imho) issue involves the many ungated or unmarked crossings that exist. What is required here, again imho, is a system which provides adequate warning of approaching trains on a continuous basis to receivers in vehicles... one reason why I supported a dedicated API for this purpose in HD FM receivers and 'talking' GPS systems... so that drivers would have warning not to go onto crossings in the first place if a train is approaching.

I sincerely wish there could be open access to metadata streams (as used for PTC) that indicate the GPS data for the front and rear of trains. Or emergency buttons on crossing equipment that, when pushed, would signal an obstacle on the tracks and halt trains safely short of that point. The difficulty in both cases is a matter of security. Railroads are already extremely reluctant to have movement data commonly distributed -- this used to be for competitive reasons, but you can easily see the fun when a gang like the old Conrail Boys whacks the signal button to stop a container train right where a krew is ready with the boltcutters and bars...

Now, I happen to dislike red-light cameras in general principle. Where that sort of tech can be GAINFULLY deployed is at crossings... ideally to get the license of drivers running gates or cutting in front of trains at the last minute. There are some fairness questions involved with enforcing laws concerning crossings with private entities (which most railroads are) but I, for one, have no hesitation in saying it benefits public safety to crack down, perhaps ruthlessly, on people trying to beat trains. Note that it is not difficult, once the cameras are in place, to have a bit of additional logic that detects vehicles stopped on crossings, and flags the appropriate authorities in appropriate ways. Having local police monitor these via some sort of distributed system, and simultaneously receiving some properly UI-enabled version of train movement data, IS a sensible way of keeping local 'law enforcement' in charge of responses to dangerous crossing situations, while keeping latency of appropriate railroad response reasonably minimized.

While numbers posted at the crossing are good, the hard truth is that some people are going to call 911 no matter what the signs say, if they even see the signs. Since the 911 computers already get information on the location of the caller in many instances, it makes sense for the 911 computer to be able to, if commanded by the operator, do the best thing it can. If that means telling the railroad police or railroad computer, so be it, but that's an extra step. You don't want calling 911 to be the wrong thing to do.

The right step is probably (if the railroad is set up for this) to have the 911 computer tell the railroad police computer about the stalled vehicle, and for the railroad computer (rather than humans) do whatever it is able to do to signal the train to slow or stop.

You do need a human in the loop to stop false alarms and kids having fun stopping trains by pushing buttons.

The theory behind the 911 system, of course, is that it's more cost effective to have a big 911 system for all emergencies than to have a bunch of independent systems. Of course, in reality, we take too much advantage of the economies of scale, and underbudget the 911 systems to the point that they can get overloaded. That is hard to fix. Logic says that adding an extra 911 staffer makes a lot more sense than a partially utilized full shift of railroad police, but we don't always get what's logical.

Of course we can have both. A number at the crossing to call, and a 911 system that is aware of what to do, even if it's just to forward the call to the number...

See also my reply below, at some greater length.

Yes, you'd have 911 involvement, and yes, it makes eminent sense to have an optimized system to reach the appropriate railroad entity in minimum time with minimal confusion. I was not trying to assert that people should NEVER call 911 in this sort of situation, only that sole reliance on 911-in-the-loop isn't the best approach -- again, imho.

I have concerns with using the 'normal' resolution of cell reception to try to determine the railroad to contact. There are numerous locations where the default resolution (by tower/channel) can't discriminate between railroads, or even lines owned by the same road. Even in those cases where 911 can call up a GPS chip in the phone -- and this poses a raft of privacy issues when enabled -- I worry that the default latency and resolution may not be high enough to discriminate between parallel mains. Furthermore, I suspect that very often the call would be made by a moving observer, whose location *by the time the 911 call is actually being processed* may be in a different cell or 'closer' to a different railroad or location; to me, this would be no different in principle from the sort of case where a 911 caller gives an address that is misunderstood as to street name or city, and responders are misdirected. This isn't of course to say 'it shouldn't be done' at all. (Naturally, HA-NDGPS solves at least some of these issues, but how likely is it that such technology will wind up in power-optimized cell phones?)

My opinion, fwiw, is that the 'person in the loop' who determines whether kids are pushing buttons, or gangs are stopping trains to rob them, or whatever, needs to be a railroad employee, not the 911 responder. That might well be different, of course, if the means of monitoring a particular crossing -- and I'm thinking particularly of whose eyes are watching the camera feed from that crossing -- maps to a local police department rather than a railroad department. Again, we have some privacy issues that might be discussed in a separate thread...

Actually most phones are configured to provide GPS location on 911 calls but not on regular calls, though more and more devices are providing location for non-emergency use. It was for emergency use that the tech first arrived in phones, and this is exactly the sort of thing it was for. While a tower triangulation may not be perfect, it's better than nothing, and it's not even that bad to signal all lines to proceed with caution approaching the crossing, which itself should be identified by such systems with good accuracy.

The main target here is commuter trains in urban areas, though of course there are grade accidents in other conditions. Commuter trains are the only ones that go high speed in urban areas, and the ones with people at risk aboard (though derailment is pretty uncommon from this I think.)

As such the location is very likely to tell you which train should slow. In addition you get different rules for crossings with long visiblity and day vs. night.

Anyway, the point is not that there aren't other ways to try to solve this -- sensors at crossings, robots etc. It is simply to propose something relatively cheap to deal with something that does happen -- people calling 911 and the 911 operator not having a very quick way to tell the train operator to slow.

Overmod - I would disagree with your statement that there is no need to involve 911.
Even if a brilliant, automated system existed, there would still some cases where, because of the distance required to stop a train, a collision is unavoidable. In such a case, I believe it is imperative to notify public safety (police, fire, etc.) as soon as possible. Also, the stalled vehicle, in addition to blocking tracks, is also blocking a public highway. This condition will also need addressed.
I do know state and county dispatchers who have encountered this scenario, were prepared, contacted the rail company responsible for the tracks and were able to make sure that train traffic was safely stopped.

I am not in any respect trying to say that there 'should not be' 911 involvement in this sort of situation. As a case in point where dialing 911 would make sense, I recently drove past a crossing where a large truck had grounded out and lifted its drive wheels off the ground. Just a mile or so later, I came across a freight moving close to track speed toward that crossing. There would be no time even to tinker with the directory system in a cell phone even if I had the railroad numbers programmed in (which I do). (I didn't dial 911 immediately; I jumped out and signalled 'washout' as the engines went by, THEN called, but yes, optimized 911 response was significant in that case, and many others as well.

The situations I (and a fairly large number of the first responders I know) disagree with involve 911 being the PRIMARY first response, particularly where large numbers of people may be trying to call to report the same incident or set of conditions. Note that the initial list of 'options' a 911 operator normally gives isn't exactly optimized for speedy response of the kind necessary to get trains stopped! It isn't difficult, of course, to say "none - there's a railroad-crossing emergency" (and the operators in my opinion usually do have prompt contact information for the local railroads), but if directed to police, for example, there will be perhaps substantial delay in getting the critical message (i.e. to get train traffic slowed or stopped in a secure and safe manner) to the right place. Yes, it's a good thing to get the police, ambulance, fire apparatus, etc. to the site of a catastrophic accident. But it's a better thing not to have the catastrophe occur in the first place... and, at least imho, it's best to keep an already-overloaded 911 system as free as possible, particularly where alternatives for more direct action exist.

I may not have noted that 'good' railroads put up a code number for the particular crossing as well as the toll-free number, in order to minimize response time and the chance for locational error. If I could make one grant proposal for maximum safety-bang-for-the-buck, it would be to get signage for every line crossing, whether vehicular or otherwise, that carried the phone and location information in easy-to-read form -- and perhaps to provide signage on mileposts for those buffoons who get stuck crossing or playing on main lines far from established crossing points.

Naturally, the 911 system should be provided with some fast lookup system correlating the codes to physical locations (and nav/GIS links to calculate optimal routes, current delays, etc. for the first responders, etc.) and the UI for that system should be made as simple and direct as possible to access. I argued long ago that 911 lines should be set up to read DTMF codes during any sort of wait, or even when operators were connected (you would NOT hear some canned 'dial 1 for police, 2 for fire, 3 for ambulance menu; it would be assumed you would know them to use them) so as to minimize latency in being connected where you want. It shouldn't be difficult to have some code that would link directly to railroad-specific connections -- and not incidentally release foreground attention of a 911 operator directly -- when calling 911 to report a railroad-related issue. [Disclaimer - I am leaving out a wide range of policy and legal issues relating to how 911 calls are monitored and recorded. I am aware of those issues, and have means of addressing them, but that is peripheral to the sense of this discussion imho]

While I have heard stories of people going on hold at 911, are there stats on how often that really happens? It should not happen of course, and presumably the main reason is people using it for non-emergencies and major events that generate huge numbers of calls. I fully agree in those circumstances that there should be an IVR, which among other things gives codes to people who declare their call involves a life-threatening emergency, and that it is unlikely that many are calling to report it. (With geolocation, the system should be able to deal with multiple calls on the same event for most events, assuring that calls from not near the event get some triage priority over all but the first few calls from near it.)

However, my main point remains the same. I was inspired to write this by the story linked above, where the people called 911 and talked to it for several minutes. Part of the fault lies with them and the 911 operator who did not establish for 5 minutes they had a stalled vehicle at a crossing. You can't fix stupid people but you can train 911 operators and you can have the geolocation computer tell the 911 operator that the caller is very close to a grade crossing and should be asked about it.

But people will call 911. At night, they may not even be able to read a sign telling how to call the railroad emergency dept. They may be too scared to punch in a number. So you need both solutions. At the very least there should be a way for the 911 operator to quickly transfer to the train police and provide the geolocation information, ideally digitally. But fewer steps is better, especially if it's just a SMOP. Many train lines are already making the GPS coordinates of their commuter trains publicly available on the web.

Add new comment