This video just appeared in my Facebook feed. I'm sure it needs more work and testing before "going live", but it could be a game changer in how goods are transported across the country. Railroads are going to want to use the same technology for their freight trains. It's a VERY impressive demonstration of computer and AI working together. No mention was made in the video whether or not the safety driver had to take over at any point.

https://www.facebook.com/UNILADTech/vid ... 136062628/

Self driving trucks have already "gone live". They have been in use in the Southwest on designated routs with a minimum wage safety driver. With all the research into self driving cars it has amazed me that there has been no mention of trains. There are a lot fewer variables on rails than on the road.

https://www.theverge.com/2019/8/15/2080 ... y-tusimple

John T. Thanks. I wasn't aware of any actual day-to-day freight deliveries with autonomous trucks. I thought it was still in the testing stage. Of course, a 2800 mile cross-country route with snow and other adverse conditions is more impressive than a dedicated Phoenix-Dallas route where winter weather hazards are minimal. I wonder what the autonomous truck would do if it lost brakes going down a long grade. Try to run off the road, run into the back of the traffic ahead, look for a runaway ramp, scan for something cheap to hit?

Here’s my prediction:

PTC was step one. Once the system is fully validated they’ll move on to making upgrades to make the engineer redundant. It will be 5-10 years. Over the road train crews will become a rare thing.

I have been a truck driver off and on since I was a PhD. student twenty years ago, on now (I got home this morning at 3:00 AM, Christmas Day). Most of us do it for the above average wages obviously, and obviously the companies would like to be able to hire someone off the street and pay them minimum wage. I don't think the public will be thrilled about it though. Anyone who can touch the pedals while seeing over the dashboard could drive a truck down the interstate if there are no other vehicles. What is tricky is making the judgement of when to turn and how sharp when going around a corner, or in a parking lot etc. The trucks currently have automatic lane departure warning and collision avoidance systems whose only purpose would seem to be to annoy the driver. The lane departure warning sounds a lot like a fart in the speaker on that side of the truck, it goes off when an on ramp ends etc. or when you are in the middle of an s-turn on a hill when the last thing you want is that fart sound distracting you. If you are using the cruise control the collision avoidance system will slow you down to the speed of a vehicle a half mile in front of you so before you know it you are going fifty MPH in a sixty, it wants us to change lanes and get in the far left lane a half mile before reaching the slower vehicle, so we can either listen to the warning with our foot on the pedal and change lanes at the last second like we are supposed to or ride the left lane and have ten people pissed at us as it takes us three miles to pass a truck. It also warns of obstructions that aren't there, the new Freightliner I was in last night mistakes an overpass on I-5 near Kelso WA for a stopped vehicle every time, flashes a red screen and will lock up the brakes if I don't have a foot on the pedal. If that is the sort of consistency this driver-less truck will demonstrate I will be very wary of driving a car near them.

After 27 years out there making trains run I can tell you that automation for railroads is a loooong way off. When I hired on in 1990 people told me trains will be automated. And there was an automated railroad in Ohio for many years, Muskingum Electric. It operated on a closely controlled right of way. On a regular railroad there are so many variables. I ran coal trains on the Escalante Western by myself for years, help was at most 40 miles away, and I wouldn't be blocking a 60-80 train a day main line! We did testing for various railroad supply companies such as Portec, and Graham White. But the one I had the most fun with over a two year period was Zeftec. We installed a prototype ECP Braking system on our train of three MK SD40-2's and 52 coal gons. There were 55 microprocessors! It was an adventure! But my main point is that every new thing takes a lot of time. And they NEVER work out the way people want them to or the economics just aren't there. Our train with ECP was a dream to operate, but as good as it was it never went anywhere really. There are things that won't ever be automated, the first and last miles, switching in yards when things change all the time, and any type of maintenance.

Autonomous vehicles are a long way away. There is a huge gap between testing and reality.

I have actually ridden a test autonomous bus in Oslo, Norway. It was not a pleasant ride. Autonomous trucks will only function in controlled environments, similar to where cruise control is already valid, on the highway between exits.

Actually, I think a better example is commercial aviation. Trucks will have a degree of automation but that will be reserved for the long haul segments and still require a human to be on the truck whenever its moving.

Automation is used by air transport pilots only while the plane is in the air, with the exception of auto-land equipped aircraft. The pilot controls the airplane during taxi and takeoff on account of the variables and risk involved at those stages. With trucks, it may well leave the driver to operate the vehicle on local roads and the automation to operate the truck on the long haul under monitoring by the driver.

One thing that has been shown by testing is that if you put a human in a vehicle and give them nothing to do except spring into action in an emergency they will find something to do. Several Tesla drivers are dead because they were letting the car drive on auto pilot and doing something else. One test showed that the more the driver trusted the automation (and the less training they had with it) the less attention they paid to driving.

I have an '18 Honda Accord that exhibits some of the same characteristics that PMC mentions re: his truck. Here in NJ, where they spray salt brine on the roads with abandon (sometimes for a heavy fog!), the vision system will sometimes track white salt streaks and give the wheel a 'lane departure' shake and light up the dashboard warning.

The collision avoidance radar, mounted under the front "bumper", will also see a slower vehicle in another lane on a curve and slow me down. There is a construction site near me with a concrete 'Jersey barrier' that narrows the one lane past the site; I get a pretty heavy brake application that I have to remember to suppress. And the radar has shut down, after driving through a puddle of slush. Wiping it off with a paper towel brought it back, of course.

The amusing thing about these systems, at least on this car, is the speed limit display. Sometimes it's misreading signs can be amusing. There are a pair of 50 MPH signs, one on each side of the road, that it has added up for me; 100 MPH. Fortunately, I know all my main routes; haven't had a speeding ticket in something like 30 years.

Yes, these self driving systems, or even 'driver assist' systems, have a LONG way to go.

I have had the opportunity to talk with a few top rail officials about their development of programs such as BNSF's predecessor to PTC, ETMS (https://www.youtube.com/watch?v=Fm7xHgpPEOc), Amtrak's systems, etc.

I have asked all of them how much further such a system would have to go to allow crewless, automated operation

They ALL looked at me as if I had three heads or spoke in Klingon or something.

They all said something to the effect of "That's not our goal here; heck, that's not even our vision." The systems in question were all seen as crew safety supplements, much as the "deadman's pedal" of old, or alerter systems that have to be acknowledged in more modern locos.

Technicians quickly rattled off a long list of why at least one crew member has to be on board: "This system can't see a tree down on the tracks; it can't see a car on the crossing or a trespasser on the track; it can't see a washout or rock fall or snowdrift; it can't see a shifted load on another train; it can't hear cars derailing behind it; on a passenger train it can't see the conductor signalling that all the passengers are aboard..........."

Now, if you're skeptical, you can read this as a well-rehearsed sop to the labor unions of "of COURSE we don't want to eliminate your jobs!" and believe that of COURSE they want to eliminate crews. But every seasoned rail industry vet I've spoken with seems utterly convinced that crewless trains will never happen save for applications like a "people mover" passenger shuttle, transit subways, a totally closed industrial loop, or the like.

Consider this: The recently-shuttered Black Mesa & Lake Powell RR was designed for automated operation, but never successfully implemented such operation. At the end, according to one trusted inside source I spoke with, the rail/power operation was paying its engineers over a thousand dollars a round trip. (Note: Some of this was apparently "political" in nature, in effect an anti-poverty program to pay at least some members of an impoverished reservation...) Given those costs, and the fact that the BM&LP was possibly the textbook example of a railroad that could be automated............

Yes, there may come a day when what is preposterously unbelievable now will become commonplace. No one, even the best visionaries, really foretold most of us walking about with GPS-equipped phones that can access the entire Internet (save China's censorship), as but ONE example. But railroads have long been conservative. And if they don't find a way to meet the challenge of the possible (crewless trucks being but one example), they will most deservedly perish.

I believe that trains and trucks will someday be fully automated with no human in them. That said, we have a long way to go before that happens, and even then I believe they will have multiple cameras on them and offices full of people to temporarily take remote control of them it they get into unusual circumstances .

Follow the money as almost always applies. Crew wages are the Class 1's second largest line operating expense just behind fuel and sometimes greater than fuel.

Autonomous mainline trains will be here FAR sooner than any of us would today consider feasible.

The sole exception will be the few mainline steamers which will continue to require a minimum 2 man crew.

IMHO-Ross Rowland

But the Tesla is not claiming to be fully autonomous. It is actually misleading to allow the drivers to give up control in that way.

Anyway, the airline example is good. I don't think in the airline example the pilots are 100% at attention ready to take over. They are drinking coffee, talking to their buddy, etc. In either example, the technology has go be good enough, and the operating environment supportive, of some level of inattention, otherwise it doesn't work.

The

We have investigated this a lot at my university, and we have an extensive autonomous subway running in Copenhagen. We had a presentation last month from one of the senior signalling persons on the suburban railway, which is installing a wireless, theoretically autonomous, system.

The problem for railways is, unless you can guarantee a sealed, protected system, it does not work. One example: our suburban system runs through areas with large deer populations. Trains hit deer all the time. An autonomous system would just keep driving for many stations with a dead deer on the coupler. Imagine the customer service and public relations impact of that.