Collision Prevention: drones, alerts, gates

Drones have been able to go 80+ MPH for years. That doesn't change anything.

Why not just use surveillance cameras. You can have them with battery power and RF to cellular communications to central dispatch who then can use the footage combined with some sort of AI that detects motion (IE something stuck on the tracks/not moving after a set time), and you could use this for all railroads. I am sure the cost of an RF based camera sending a cell signal (since most of the US has 4G service now) is much lower than a drone.

miamicanes wrote:Here's a better idea: a robot running on the tracks *just* far enough ahead of the train to trigger auto-braking, but not entering the crossing until 5 seconds after the gates are down.

....

If you do that, you will see a huge upturn in auto/rail collisions with the 'robot' hitting go-around cars. If a certain number of cars each year just don't see a full-sized locomotive, you can bet a much larger number of them won't see a robot preceding it. Add to that the likelihood of just such a collision resulting in a disabled wreck on the tracks...

The thread's not about Predators- I think they're talking about Off-The-Shelf camera drones, like the ones you can buy in any hobby shop.

Ryand-Smith wrote:Why not just use surveillance cameras. You can have them with battery power and RF to cellular communications to central dispatch who then can use the footage combined with some sort of AI that detects motion (IE something stuck on the tracks/not moving after a set time), and you could use this for all railroads. I am sure the cost of an RF based camera sending a cell signal (since most of the US has 4G service now) is much lower than a drone.

This.

Oh sure. And who exactly is going to monitor 100,000 security cameras all day and all night? That sounds neat, but it is technologically, logistically, and practically ludicrous. People are going to drive their cars in front of moving trains. There is nothing that will stop that or prevent the collisions. All this drone and camera stuff is absurd.

MCL1981 wrote:Oh sure. And who exactly is going to monitor 100,000 security cameras all day and all night? That sounds neat, but it is technologically, logistically, and practically ludicrous. People are going to drive their cars in front of moving trains. There is nothing that will stop that or prevent the collisions. All this drone and camera stuff is absurd.

Again, I work with modern AI driven tech. It is a fairly easy problem to say "Stuck" since you only need the AI to sense if an object does not move after a small period, so you don't have a human operator unless the AI signals dispatch to let the human know to put the camera on view.

(tbf I remember someone in here calculating that it is honestly cheaper to remove 95% of grade crossings (and every major passenger railroad should have 0 so your NEC family rails), than to do this so you aren't wrong)

MCL1981 wrote: People are going to drive their cars in front of moving trains. There is nothing that will stop that or prevent the collisions.

This extra air surveillance would be really helpful in many problem-prone places such as rock slide or flood-prone areas. I can see that would be a good place to try it out.

As far as cars and trucks stuck on crossings, why the hell don't all crossings just have typical detection loops in the pavement? They can't be very expensive to install, they are everywhere now. If the loop detects a vehicle on the crossing there would be some kind of notification to the train. A simple flashing warning strobe might work (placed before the crossing if on a curve). Seems really simple yet I don't think it has ever been done.

Most collisions aren't vehicles stuck on the tracks. They're dumbasses that just drive around the gates into the path of train. They cannot be detected, and there is no stopping a train in that amount of time. Exercise in futility.

State and federal law should simply prohibit civil lawsuits when the crossing protection equipment was found to be working properly, and where the engineer was found to have used the horn and lights as required. If someone is stupid enough to drive their car through crossing gates and get hit by a train, frankly the gene pool is better off without them.

MCL1981 wrote:Most collisions aren't vehicles stuck on the tracks. They're dumbasses that just drive around the gates into the path of train. They connect be detected, and there is no stopping a train in that amount of time. Exercise in futility.

State and federal law should simply prohibit civil lawsuits when the crossing protection equipment was found to be working properly, and where the engineer was found to have used the horn and lights as required. If someone is stupid enough to drive their car through crossing gates and get hit by a train, frankly the gene pool is better off without them.

I'd rather the judicial branch just got more funding for more judges and clerks, so suits could be dismissed much more quickly, but I see your point too.

Also, we're overdue for many more quad gates with median protection, more extended gates for rural roads, and more protected crossings altogether. These are more expensive than drones, but a drone doesn't provide positive protection against incursions.

Needeth we another "motorman/locomotive engineer" set-to around here?

David Benton wrote:

Ryand-Smith wrote:Why not just use surveillance cameras. You can have them with battery power and RF to cellular communications to central dispatch who then can use the footage combined with some sort of AI that detects motion (IE something stuck on the tracks/not moving after a set time), and you could use this for all railroads. I am sure the cost of an RF based camera sending a cell signal (since most of the US has 4G service now) is much lower than a drone.

This.

Heartily agreed. I can't believe this discussion has gone on for so long.

Tadman wrote:

David Benton wrote:

Ryand-Smith wrote:Why not just use surveillance cameras. You can have them with battery power and RF to cellular communications to central dispatch who then can use the footage combined with some sort of AI that detects motion (IE something stuck on the tracks/not moving after a set time), and you could use this for all railroads. I am sure the cost of an RF based camera sending a cell signal (since most of the US has 4G service now) is much lower than a drone.

This.

Heartily agreed. I can't believe this discussion has gone on for so long.

• Raspberry Pi Zero: $5
• Name-brand SD card: $15
• Pi Official Camera: $25
• USB LTE modem: $50 (from looking at Amazon)
• Battery HAT for Pi: $65 (PiJuice HAT from PiSupply)

Total: $160/monitor. Depending on a different battery HAT and using a different way of communications, this can go to $100/monitor, just with off-the-shelf parts.

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.

MCL1981 wrote: State and federal law should simply prohibit civil lawsuits when the crossing protection equipment was found to be working properly, and where the engineer was found to have used the horn and lights as required. .

Oh yeah, because there's no way that won't be abused.

Nothing out there now provides an accurate time stamp, or a traceable log of events anyway.

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.

4G is cheap, and the data rates wouldn't be high most of the time anyway. Who the hell is going to watch cameras? It's all about vision systems these days. Vision systems watching every camera, bouncing the suspect ones to the dispatcher and maybe even setting the approaching signals to restricting if it feels things are REALLY suspect and there's not enough time for human intervention.

This is all off the shelf tech at this point. Automated vision systems are used everywhere to inspect things. Cellular is cheap and plentiful. VPNs are standard features in cell modems.

It's not 1975 anymore.