Collision Prevention: drones, alerts, gates

MCL1981 wrote:Plus the cost of 4G service to each unit. Plus the cost of a few thousand new employees to monitor these useless cameras all day. Not a single grade crossing collisions will be prevented. Not a single life will be saved. And it will simply increase the railroad's liability in a lawsuit. It's a horrible idea that changes nothing.

Existing software (called "motion") can cut down the service cost (only sending alerts to the railroad dispatchers should certain thresholds get met, like a person or a car suddenly go on the tracks and stay there for more than a minute). Having a fiber back-haul to a relay system can also cut that down as well.

Dispatchers (yes, they'll have to hire some more) will have to react on the alerts, and having a system to "dial in" and get the latest imagery also will help. Some of this can be mitigated by having the cameras feed into the dispatch computers, and the computers would drop the signal to Restrict should an alert happen.

Will it prevent all the collisions? Nope, too many variables, including faulty PTC, signaling, crossarm failure, stupidity of humans, phase of moon, etc. But given a well researched specification, I can do it with off-the-shelf hardware and open source software, and cut that collision rate down.

I will say one thing though that will sink the "cameras along the line" idea. You need one every mile of track due to two insummountable issues: Camera resolution and the curvature of the earth. Consider that in a cost analysis.

They only need to look at cameras just in advance of a train arrival at that spot. You don't need a legion of watchers .

MCL1981 wrote:Plus the cost of 4G service to each unit. Plus the cost of a few thousand new employees to monitor these useless cameras all day. Not a single grade crossing collisions will be prevented. Not a single life will be saved. And it will simply increase the railroad's liability in a lawsuit. It's a horrible idea that changes nothing.

As noted, automated vision systems are *extremely* advanced. Watching "if the gates are down, have the tracks cleared" is a simple enough question that a smartphone-level processor could answer it with a higher reliability than a human (who might get distracted, fall asleep, etc...). There is a small initial cost - setting up artificial neural networks for this task - which is about as difficult as might be assigned in an introductory university course on machine learning. This system could be hooked directly into the PTC, so that trains would automatically come to a stop unless the cameras confirmed it was safe to pass.

I'm not sure how the logistics of this are escaping you all.

Morons drive their vehicles around gates seconds before a train hits them. Putting up hundreds of millions of dollars worth of cameras (or drones), infrastructure, and personnel is not and cannot prevent that collision from happening. That collision is going to happen, period. That constitutes 86% of public grade crossing collisions in 2017. This fancy and pointless system that will never work is useless for 86% of the collisions. The only way to prevent that grade crossing collision from happening, is to remove the grade crossing.

In 2017, there were 464 collisions at public grade crossings. Only 50 were the result a vehicle being stuck or stalled on the crossing well in advance of the approaching train. Your hundreds of millions of dollars worth of cameras (or drones), infrastructure, additional personnel, R&D, testing, and training may prevent a some of those 50 collisions. So you want to prevent a small percentage, of an already small percentage of collisions. Pointless exercise in futility. That money could be spent on far more effective tactics.

MCL1981 wrote: I'm not sure how the logistics of this are escaping you all. Morons drive their vehicles around gates seconds before a train hits them.

Grade crossings should be upgraded with medians and guardrails and four-way gates. They are too easy to drive around. In most cases freight trains are getting longer and longer and the wait at a crossing is taking a lot longer now-a-days and people are more impatient than ever.

Perhaps it's also time to consider a more substantial gate like this? At least at "high-speed" crossings.

That is an example of money being spent on more effective tactics!

Of course my personal opinion is a little more harsh. Save their money. Thinning of the herd has been working effectively in nature for millions of years. Why stop it.

MCL1981 wrote:I'm not sure how the logistics of this are escaping you all.

Morons drive their vehicles around gates seconds before a train hits them. Putting up hundreds of millions of dollars worth of cameras (or drones), infrastructure, and personnel is not and cannot prevent that collision from happening. That collision is going to happen, period. That constitutes 86% of public grade crossing collisions in 2017. This fancy and pointless system that will never work is useless for 86% of the collisions. The only way to prevent that grade crossing collision from happening, is to remove the grade crossing.

In 2017, there were 464 collisions at public grade crossings. Only 50 were the result a vehicle being stuck or stalled on the crossing well in advance of the approaching train. Your hundreds of millions of dollars worth of cameras (or drones), infrastructure, additional personnel, R&D, testing, and training may prevent a some of those 50 collisions. So you want to prevent a small percentage, of an already small percentage of collisions. Pointless exercise in futility. That money could be spent on far more effective tactics.

Fully agreed.

MCL1981 wrote:That is an example of money being spent on more effective tactics!

Of course my personal opinion is a little more harsh. Save their money. Thinning of the herd has been working effectively in nature for millions of years. Why stop it.

I see where you’re coming from, but nope, considering that one impatient driver can kill a school bus full of kids! They haven’t deserved thinning.

Ok, good point.

If the issue is grade crossing safety (as opposed to rocks on the track, barricades built by terrorists or stupid snowboarders building jumps (I have called the cops about this once)) then some study and sorting can help.

1. Watch the busier crossings to see where lots of people routinely race the train. Then install license-plate cameras to write tickets at those crossings, with a big sign to that effect, same as they do at tollbooths. Either you solve the problem or you make money that you can then invest in 4-quadrant gates, upgrading lights-only or unsignalled crossings, etc. If the software and wiring diagrams to photograph all cars passing x point when the gates are down/signals are activated does not exist yet, it will not be hard to write. And since someone running the gates is by definition also in the wrong lane, it would be difficult for them to go to court and claim the camera was wrong/they were already crossing the tracks when the signals began. Yes, in most places this would require cooperation with a host railroad and a local police force, but it would help them, too.

2. Do the traditional things to prevent crossing accidents: quad gates, medians, upgrade lights to gates and crossbucks to lights.

3. Identify crossings with poorly timed or frequently malfunctioning signals and get them adjusted/fixed, or if both slow and fast trains use a crossing upgrade to the predicting software type that always drops the gates the same number of seconds before the train reaches the street. Same for signals that are frequently turned on by the rear of a train that has cleared the road but not the crossing circuitry and stopped for a meet or some other reason. I know a crossing (Hulman St. at CSX in Terre Haute, IN) that is frequently not quite blocked by southbound trains waiting for meets; the gates go down, the train waits five or ten minutes, and many people snake their way around the gates. Very bad training/conditioning for drivers. Find an electronic or operational way to get those signals off.

4. Study whether it would help to speed up freight trains in areas with busy crossings. Some cities (incl I think Terre Haute) have speed limit ordinances for trains because they think slower is safer; maybe so, but if the limit is thirty but the track could handle forty by FRA rules, a simple ordinance could reduce time that crossings are blocked by a quarter.

Agreed with that too, Arbor. Also:

5. When crossings are consistently blocked or partially obstructed, use public money & matching private money to fund aiding extensions, traffic flow improvements, and other railroad infrastructure that can help relieve road-rail interactions.

mtuandrew wrote:Agreed with that too, Arbor. Also:

5. When crossings are consistently blocked or partially obstructed, use public money & matching private money to fund aiding extensions, traffic flow improvements, and other railroad infrastructure that can help relieve road-rail interactions.

Which would be a really good idea until someone asks, who’s going to pay for it?
At street-railroad grade crossings, who is the responsible organization financially - railroads or streets?
In most cases, it is the streets-highway departments.

Legally yes, I think you’re right, Ron. I’m suggesting that Congress provides for a USDOT fund to allow them and state DOTs offer matching funding for track improvements (rearranging sidings, etc.), to increase throughput and safety for both road and rail modes. Perhaps any import duties and tariffs can be shunted to such a fund.

Who is going to pay?
1. Who was going to pay for the drones and their monitoring?
2. Railroads, if the improvements can get them higher speeds, greater capacity, reduced delays/lawsuits from crossing accidents.
3. Communities that own the roads.
4. State and federal, as for other big infrastructure projects.

Example: overpass over CSX in Terre Haute, IN, being built by city with state and maybe federal money and I believe with help from the railroad, who will get a 3-track passing place (partly in a four-five track yard) without grade crossings for a mile and a half, and with only one secondary street grade crossing for about 2.5 miles. So the city gets a through route that will never be blocked, the railroad and the city eliminate possible grade-crossing accidents, and the railroad gets the ability to hold trains without blocking traffic (or to hold longer trains that before without blocking traffic -- not all trains currently block traffic when held at this site, but many do).

Not every crossing elimination or improvement has such mutual benefits, but many do and could therefore be done collectively if there was just a bit more public money for the public/street share.

mtuandrew wrote:I see where you’re coming from, but nope, considering that one impatient driver can kill a school bus full of kids! They haven’t deserved thinning.

I wonder whether there is a practice (or if one could be instituted) of permanently displaying, on the dashboard of every school bus in plain view of the driver, the legend
THIS VEHICLE IS xx FEET LONG
One tragic collision on the UP NW line in Chicago some years ago resulted partly from the fact that the driver was operating a longer bus than the one usually assigned and may have believed, when she pulled across the tracks and stopped behind the stalled traffic ahead, that the bus was clear of the crossing. It wasn't, and I think about six high-school students in the rear of the bus were killed.

I like it, ExCon. If rental trucks can have their height on the dashboard, school buses, other buses, and non-trailer-trucks can have their length there. Cheap and easy.