Hacker News Reader

Comment by EvanAnderson | Hacker News Reader

There's also a trial of "platooning" of driverless trucks on I-70 in Ohio and Indiana: https://www.forbes.com/sites/richardbishop1/2025/04/24/ease-...

A "drivered" lead truck is leading one or more driverless trucks in this case.

I drive the stretch of highway these trucks are on fairly regularly. I don't know that I've seen a group of them yet but I'm keeping my eye out.

I'm probably just showing my age, but I like the idea of a "drivered" truck leading driverless trucks versus a completely autonomous system. It's similar to my attitude on crewed spaceflight-- I like the idea of the ingenuity and capacity for independent thought supervising an automated systems, versus autonomous automated systems.

View on HN

next level would be to hook these "platoons" together physically and then centralize the propulsion in a super efficient package. And then we could move them off the highways and onto specialized "tracks" that guarantee they don't deviate from the planned routes.

speculative, alien technology, admittedly, but some day our scientists will figure it out i bet!

Hmm. That sounds like it might lead to a monopolization of these tracks, as networks consolidate, and then ultimately evolve into a stagnant industry more focused on cutting costs than innovating...

I suspect that's why a competent government might invest more in these tracks as they might contribute a lot to delivery infrastructure. Especially in a consumer-based economy where shipping goods is important.

Let me know where you can find a competent government not currently trying to burn down its own consumer-based economy with tariffs and xenophobia and which doesn't consider trains a communist plot. And whether they take American passports.

I'm not sure governments learn by seeing the mistakes of other governments. But consecutive New Zealand governments have made some big mistakes. Capitalist party sold the national rail to an Australian company because rail was inefficient and costing a ton. Privately run cut back staff significantly. A few years later the story was that that was failing privately run, so the socialist party bought it back at exhorbitant cost. Now current governmentis trying to sort out a massive budget blow-out to get "rail" ferries between North and South Islands working (roll-on roll-off). The private ferry (Bluebridge) is doing okay I think.

I believe the lesson is that rail sucks here, economically speaking (whether privately run or publicly run). Unfortunately old voters love rail so the politicians pander to them.

Dumb American, so I could be totally off base. But New Zealand appears small enough on the map that just doing a shit ton of barges along defined coastal water routes makes a lot more sense than trains. In the US trains could be efficient, but we are also like 2,500 miles coast to coast.

The problem with water travel in New Zealand is that weather often closes ports.

But it is true that NZ lacks the scale that many existing rail systems rely on, and also lacks high-speed roads (a good rule of thumb for driving anywhere in NZ is: look at how many miles it is on the map, calculate how fast you could go that far in America, then double the time).

New Zealand doesn't have inland waterways the way the US does so barges don't seem sufficient. They did actually get a fancy self-propelled barge for use transporting minerals on the West Coast of the South Island last year and it got pushed aground by a storm in its first couple weeks of service. It's been sitting in port in Nelson ever since being refloated.

https://maritime-executive.com/article/barge-to-innovate-new...

The problem about New Zealand is that they hate investing in infrastructure and everyone loves the kiwi myth of the number 8 wire which means they just build provisional solution on top of provisional solution. The example of the ferries is a good one, the cost blow out was because they needed to build a new dock in Wellington. So instead of just living with the fact that yes building infrastructure as become expensive in a country which has seen a massive building boom (like many other places), the new Conservative government decides to cancel the ferry order without any alternative in place, and now the most important connection of the country is running on failing ferries which are more than 25 years old (there have been several incidents in the last couple of years where they were drifting powerless), and (especially considering the massive penalty for cancelling the contract) the new ferries will be just as expensive as the cancelled ones just being much smaller (the old contract was actually signed at quite advantageous economic times) and arriving much later.

This is certainly not an example of rail not working. Trucks on the road are a massive disaster as well. Most roads are single lane, so it's not uncommon to be stuck behind trucks for long stretches, this is especially annoying considering that many roads are very windy and truck speed up when coming to passing lanes.

Trains are as popular as ever almost everywhere in Europe.

Rail freight's share of shipping has been on a steep decline for decades in Europe:

> Freight rail’s modal share has been in decline across Europe, both in terms of market share and the profitability of major operators. In France for instance, modal share declined by 50 percent, from around 30 percent in the 1980s to 15 percent today. By contrast, road transport has been steadily increasing. In 1980, less than 50 percent of goods were transported by road. This rose to more than 75 percent by 2018.

https://www.mckinsey.com/industries/infrastructure/our-insig...

By contrast, for most of that same period (1980-today) rail freight was growing its share in the US, though in the past decade or so growth has favored trucking.

Trains move as much freight as ever in the USA.

Not to take away from that fact, but the share of train-moved freight (over time) is more interesting. It probably highlights the complications of scaling train networks easily.

Switzerland?

Austria, France and Italy are currently copying Switzerland in building base tunnels [1] to mostly carry rail freight.

[1] https://en.wikipedia.org/wiki/Base_tunnel

US is already the largest shipper of freight by train in terms of shipped tonnage and also total tracks.

https://en.wikipedia.org/wiki/Rail_transportation_in_the_Uni...

Not a very interesting data point since it doesn’t account for nation area, utilization, etc.

Don't forget geography, and by extension type of freight.

The United States doesn't have a lot of navigable waterways, and none go east-west. This means that any bulk goods you want to transport have to go via rail. There is no other option. As a result the railway network has been optimized to do bulk transport at the lowest price possible, even if it means sacrificing speed and quality. After all, coal doesn't care if it sits in a siding for several days.

It's true that the US is shipping an awful lot of tonnes via rail, but this doesn't say anything about its ability to ship regular goods. Getting a single wagon from point to point in a timely manner is significantly more complicated than doing the same with an entire train.

https://commons.wikimedia.org/wiki/File:Freight_Shipping_in_...

Here’s tonnage miles by class.

For comparison. There is only 20,000 miles of railway that Amtrak operates on. Thats a fraction of passenger railway compared to China and EU.

dammit

Ah, you're thinking of the world famous Las Vegas Hyperloop!

Sounds like something that would require a lot of infrastructure that wouldn't make a lot of sense to build in the most rural areas of a country, unlike roads which can be quite cost-effective even without towns nearby.

Problem is those railroad cars usually don't share origin and destination. And spending a day or two on the switchyard for every couple of hours of actual travel isn't what customers want.

The beaty of platooning is that trucks can join and leave while traveling at speed, all they need is a free lane on the side.

And if you really want to tease out all the efficiencies of rail, without suffering the drawbacks, you make electrified lanes that also have embedded tracks (like those of a tramway, but designed for higher load) and make the trucks multimodals that can switch while moving. Actually just those trucks that would see an economic advantage, trucks designed for light, high volume loads would likely stick to road wheels only.

A setup like this could actually be super economical compared to two because a rail network needs huge separation between trains because rails are terrible for braking fast. Multimodals on the other hand could be designed to be able to switch to their road wheels in an emergency. And a rail network needs an ungodly amount of nines in terms of reliability and rarity of maintenance, because there is no plan B network. Reliability nines are expensive. Multimodals on the other hand could easily be diverted to a dirt road for a bit when the main lanes are defective or going through maintenance. Where they would be running on the same batteries they carry anyways, for the last couple of miles on regular, non-electrified roads.

switchyard

Why not just use container cranes like docks? There's no reason they won't work inland.

And indeed they exist every 200 miles or so.

Would that be faster though? You still have to stop the entire train just to get one unit in or out. And if it's more units, chances are that parallelization hits a capacity bottleneck because of the inherent cost trade-off between having lots of cranes on standby vs just waiting a little longer in the rare moments of action. Are container cranes used for routing within a rail network at all? Might well be that they are only ever used for mode change at the edge, despite having been mature and available technology for decades.

So why aren't trains automated yet? I mean, beside the inter-airport shuttles. Seems like a much more constrained set of parameters to deal with. I'd think this would be a good "first" project to tackle for an AV company, even if the economics of replacing locomotive drivers/engineers weren't super compelling. Is it an insurance thing? Because a train wreck could be much more disastrous and costly? Or is it a union thing? Others?

> So why aren't trains automated yet?

Why? The labor of a conductor is a fraction of the total cost to move the cargo, to the point it’s a rounding error.

And the job creates a sense of community too.

At some point it’ll probably be automated, but yeah, what’s the point.

And it means someone on site to react sensibly when things go wrong.

Think of the Miracle on the Hudson. The pilot said "This is the captain. Brace for impact." The flight attendants picked up on that and immediately began directing the passengers on what "brace for impact" means. Without that how many would have reacted properly (the info is on the safety card, but how many have read it well enough??), how many would be asking for information, how many would just generally be panicking?

If a freight train is unstaffed, who would brace for impact?

Anyway, labor costs to have a couple engineers aboard is much less than if the load was travelling by truck, even if the trucks have no driver in them, they're likely supervised at a ratio with more humans than it takes to operate a long freight train.

That was a general example, not a specific one.

And in my experience conductors are passenger trains. Freight might have a driver, but no conductor.

Didn't we almost see a national rail strike 2 years ago? It seems like reducing that risk would be an incentive enough to make the transition, even if overall costs are low.

Automating the conductor job doesn’t necessarily mean there won’t be anybody around to strike, I think.

This. Most of the jobs are in infrastructure and servicing (loading, maintenance etc). Those are using as much automatization as possible already.

> beside the inter-airport shuttles

Singapore and Sydney would like a word.

Nuremberg(partially), Hamburg(partially, soon), too.

https://www.grandparisexpress.fr/new-metro

Lots of trains are.

https://en.m.wikipedia.org/wiki/List_of_driverless_train_sys...

Almost all of the currently automated rail systems in the world have two things:

1. Full grade separation with no level crossings (often including platform edge doors along any station platforms the trains move through too)

2. Full fencing along the corridor.

Those are two things that isn’t true for almost any freight line.

Not to say it’s not possible with extremely extensive sensor and decision making systems but we’re probably not really there yet.

Wouldn’t it be great to enable electric cars to go long distances with platoons? Like a bus system for electric cars to go between cities and charge while you drive. The ‘bus’ is basically a big efficient generator that can charge up the cars behind it. A lot like mid flight refueling for fighter jets.

Generator powered by what? If its a diesel powered generator, that's even less efficient and more polluting than just using diesel trucks in the first place.

Hydrogen

As I understand the benefits of electric cars themselves, the burning of hydrocarbons at scale is more efficient than piecemeal. So if the ‘bus’ contained a turbine, there would probably be net benefit. Or a nuclear bus!

Plus, even if it’s equal to the present state of affairs, we need a practical road to electrification of transport. I’m sure there are many cheaper options, it’s just fun to think about the infrastructure options.

The problem with railroad is exactly the other side of this: that you can't trivially and automatically detach a single car at any point. I know businesses that literally have an old railroad branch line right up to their door, but they can't actually order anything meaningful via rail.

The existing rail system has at least the following constraints:

* Uses steel-on-steel friction, rather than rubber-on-rock. Cars that can do both exist but are rare.

* Can only travel on the specially-prepared rails, not installed at the last-mile, related to the next point.

* Poor cornering and elevation changing.

* Difficulty in changing speeds (over a thousand meters and over a minute, compare to about a hundred meters and less than ten seconds for road vehicles)

* Very limited lanes, usually no passing. Track reservations will be voided if you aren't exactly on schedule.

* Almost all of the "intelligence" (both computer and human) is at one or both ends; the cars in the middle are all "dumb".

Which of these can reasonably be changed?

My previous company, Parallel Systems is working on this. The solution looks a lot like trucks on rail: individual locomotion, so you get the flexibility of trucking, but the energy efficiency and automation ease of rail. With modern braking, you can stop a Parallel vehicle about as fast as a truck. There's a ton of rail that's underutilized, particularly in Metro areas, so you almost get a fresh highway for free. Vehicles can platoon or separate at will.

Always upvote electric trains!

I’m not seeing much rail that is under utilized out here int he west, if anything it is over subscribed. The energy savings in having one or two diesel electrics pull a long chain of cars is substantial, I get the feeling that if you just put trucks on rails you’d lose a lot of that.

You also need to be more conservative with elevation changes, right of ways, and turning radiuses, so lots of tunneling and viaducting ala the Chinese HSR network. You are still probably going to have a lot of roads since rubber works better for going up and down, twisty turns, and can deal last mile stuff flexibly. Not everyplace is going to be as simple as the island of sodor.

most of Europe north of the alps is fairly flat in elevation. Railways are oversubscribed because they are not expanded enough, defunded if you like, to expand and fund the highway network.

It is stupid to travel thousands of kilometers by trucks. It's inefficient, expensive and nit scalable. Currently we use mostly huge container ships and empty them with trucks, absolutely insane. There's quite a bit of river boat activity in Europe luckily, which is great if electrified. Though the sorry state of the railways is just human-made and not didtated by economics.

>the sorry state of the railways is just human-made and not dictated by economics

I'm not sure about the world in general but I've followed the HS2 line in the UK and the problems are pretty much economic. We have a couple of north south lines along the UK but they are basically at capacity so the idea was to build another and make it high speed but it's proved incredibly expensive, partly due to property ownership and housing along the line and partly due to environmental regulations leading to £100m shelters to avoid disturbing a few bats. (bat thing https://www.theguardian.com/uk-news/2025/jan/25/environment-...)

Estimates for building it have gone from £38bn in 2011 to £136bn in 2023. They are still building some of it but brexit britain doesn't have that many spare £136bns lying around. https://www.economicshelp.org/blog/3088/economics/pros-and-c...

Yeah the big infra projects in the West somehow became like this. It's a very different story in Asia though.

They can all be changed, but then the capital costs for rail infrastructure would balloon and diminish the 3-5x cost advantage rail has over trucking. The modern rail business is structured (physically and, especially, financially) to squeeze every last cent out of existing infrastructure with minimal investment. I doubt that will change significantly anytime soon. The public road infrastructure subsidizes trucking (even after factoring in use and fuel taxes), and perhaps more importantly trucking companies don't have to contend with legal hurdles like land acquisition--the government handles that. (Land acquisition delays have been the biggest barrier for CAHSR.)

>The problem with railroad is exactly the other side of this: that you can't trivially and automatically detach a single car at any point. I know businesses that literally have an old railroad branch line right up to their door, but they can't actually order anything meaningful via rail.

Because companies have invested in JIT because low cost of trucking, an industry which is heavily subsidized (https://www.cbo.gov/publication/50049), allows it.

If trucking costs reflected their true cost, railroad would look a ton more attractive along with not doing JIT.

The bigger issue is that roads are immensely subsidised compared to rail. Even on private roads that are completely covered by usage charges, trucks pay maybe a factor of 2-10 more than cars, however they cause about 40000 times more wear (let's even disregard how they also cause disproportionately more traffic...). Obviously this calculation becomes even worse if we consider that most roads are subsidised by all tax payers. Trucks would be much less profitable (especially for long haul traffic) if they would have to cover their actual cost to society. But they are just another example of privatising profits and socialising the costs.

Are you just talking about the USA? I would bet that rail in Europe and Eastern Asia is also pretty subsidized, as much as or even more so than roads.

Rather the opposite in Europe, actually.

The European rail network is strongly liberalized, starting with the First Railway Directive 91/440/EEC. The physical infrastructure now usually falls under a state-owned enterprise, which provides access to any company wanting to use it, at a fair and non-discriminatory pricing. In practice this means that the usage fees have to pay for all upkeep and maintenance, with occasional cash injections from the government for things like constructing new railway lines. Importantly, this infrastructure company is expected to either have a neutral result, or run a profit: the railway users have to pay the true costs of operating the railway.

The roads are a different story altogether. They are owned, constructed, and maintained directly by the government itself. Road upkeep comes out of the general budget - just like education or defense. Road users pay via a vehicle tax and/or fuel tax, but there is zero expectation that those taxes pay for the full cost of the road network. After all, you profit from the presence of a road network - even if you don't drive a car yourself.

France has fairly high tolls on roads and Switzerland has that sticker you have to buy to use the expressways.

Japanese rail companies must get some sort of subsidy on tracks, the passenger rail system runs at a loss and only turns a profit on renting space in train stations. Roads are also heavily tolled (similar in China).

> France has fairly high tolls on roads and Switzerland has that sticker you have to buy to use the expressways.

France has tolls, which pay for 90 roads (granted, big ones), out of thousands.

that 40000 figure is the product of one study. I have always been suspicious of it and wished people could cite more than one study written 40+ years ago

And states like Rhode Island trying to actually charge truckers for those damages keep having their efforts struck down in the courts.

You can. It’s called multimodal and basically is just picking a shipping container off a train car with a crane and putting it on a truck platform or a ship. 1 platform supports all modes.

This works great for finished consumer goods, but bulk cargo that makes up a significant portion of both rail and truck traffic like grain, liquid products (crude oil, gasoline, vinyl chloride), ore, etc. have very specialized transports that don't work well with the existing multimodal system.

Most of those have their own specialized transport systems that work well already though.

Intramotev is working on autonomous rail car retrofits to make the cars in the middle move on their own to their destinations.

> I know businesses that literally have an old railroad branch line right up to their door, but they can't actually order anything meaningful via rail.

A century ago they would have found it at lot easier: if you wanted to send just a few cars to the back of a factory, many rail operators would have found a way to make it happen, and at a reasonable price too - now they’ll tell you the service is uneconomic to provide, unless you charter an entire locomotive for $$$$

The problem with rail is that we allowed the national rail network to be privatized. Maybe we’ll see some privatized lanes on interstate highways in the next 15 years.

railroad maintenance is also incredibly expensive.

Especially compared to when you can offload the road maintenance to the state / rest of the population.

As repairing roads costs so much, it annoys me that we don't make some attempt at penalising the heaviest vehicles. As road damage is roughly proportional to the fourth power of weight, it would make a lot of sense to prevent logistics companies from just using the largest vehicles that they can - it should work out costing less if loads are broken up and delivered via multiple, lighter vehicles. However, as the public bears the cost, it means that logistics companies will just look at their expenses (drivers' wages, fuel etc).

And people could just vote for things to be better!

It's fascinating the frequency with which techbros invent "trains but worse".

Google Australian road trains.

Jokes aside, Germany trialed a hybrid catenary truck system which was really neat.

Diesel to get you from the warehouse to the highway, extend your pantograph and boom, you're electric in the right lane (outside lane) all the way to your exit.

Never looked into how it penciled out but I always thought that was a great idea. Seems like there is a path to add in driverless aspects to the system as well given you have contact with a wire the whole time.

It penciled out by striking it through. Meaning they all have been built back, and aren't in operation anymore, with no further experiments planned.

Also disinterest by truck manufacturers.

> next level would be to hook these "platoons" together physically and then centralize the propulsion in a super efficient package.

You don't need the tracks, and its called a 'road train'

https://en.wikipedia.org/wiki/Road_train

Yeah! Then those same engineers will notice how much single point of failure that model leads to, and how rigid that is. So they will be looking to leverage the existing much denser, much flexible, more connected and redundant network that exists alongside those rare tracks.

It’s almost like this evolution did not happen before.

Imagine having rails instead of roads, and roads instead of rails. The same argument would apply, only roads would lead to less efficiency overall.

turns out, the US has already done this at a scale larger than anyone else https://en.m.wikipedia.org/wiki/List_of_countries_by_rail_tr...

Trains have been with us for over a century and aren't going to suddenly change on their own to solve all transport problems. Self driving tech may help make them work better by handling the station to end location bit cheaper or more efficiently than human workers.

This reminds me of an oft recommended book "Digital Apollo". One of the driving topics is the human interaction component and the difference in designing a fully automated system versus one that is designed with an operator that can intervene. If I recall correctly, the book presents a dichotomy between the rocketeers and pilots (automate entirely and strap people on for a ride vs design a system controlled by a human).

I think they both have their place, but I think acknowledging it as a system design choice is so helpful even in basic business processes (how will I handle exceptions, how will the person remember to handle a rare exception).

I find myself thinking of this problem frequently. We have lots of modern words for it like observability but I think that removes one a bit from the actual problem.

> A "drivered" lead truck is leading one or more driverless trucks in this case

My bet is this goes nowhere. It’s a horseless carriage that doesn’t have enough time to pay itself back versus fully-automated platoons with remote back-up.

It seems like there's an aerodynamic advantage to the platoons. Placing an autonomous truck in the lead of one of these platoons, down the road, seems like a reasonable "upgrade" strategy.

> Placing an autonomous truck in the lead of one of these platoons, down the road, seems like a reasonable "upgrade" strategy

Sure. I just don’t see the time-to-market advantage of starting with a human lead outpacing the core technology advantage of being fully autonomous from the get go. (Counterpoint: Waymo using Uber to manage the front end in Atlanta.)

But the company that figures out platooning can also work on self-driving, or can partner with or sell itself to the company that does figure out self-driving.

Working on platooning - assuming it's a more or less viable business idea on its own - doesn't need to be a distraction that prevents anyone from developing fully autonomous trucks.

Right up until there's an accident and the lead truck suddenly has to slam on the brakes. Depending on the length of the 'platoon' and traffic density, that could get really messy really fast.

All the trucks should have their own set of brakes, so the stopping distance should be similar for all of them, right?

Unless the front truck is actually hitting something, causing it to slow down more than the following trucks (and then quickly turn into a pancake).

If we can fly a drone halfway across the world, why can’t we have human augmentation for self driving trucks. Have a human, rested, and alert monitoring several platoons back at HQ. The only reason you want a person on the train is security and paperwork.

We can't reliably fly a drone halfway across the world. The mishap rates for remote piloted aircraft are terrible compared to similar manned aircraft. And for trucks, we can't even get fast, reliable cellular or satellite data on a lot of roads (including tunnels).

I'll bet one more cynical.

The teamsters, the CDL mills, the driver trade groups, all the people who've joined hands with the hand wringing types to prevent modernization of our trucking regulations will see the writing on the wall, pull out all the stops, lobby to allow human steering wheel holders to work to their full potential and full automation will stagnate for 20+yr while tech is instead deployed to solve other nuisance problems that limit truck sizes and productivity.

Someone should develop an ad-hoc platooning network for truckers. Install a platooning cruise control package on your rig and then get an alert on your dash when another truck in the network is in your vicinity looking to platoon. Lock in the cruise control behind the lead vehicle, and the app automatically calculates the fuel savings and divides the savings equally between the operators.

Edit: Various sources say that platooning can cut fuel use by 4-15%.

> There's also a trial of "platooning"

That's been tried before. See Demo 97, when CALTRANS had a demo of self-driving.[1] Worked OK, but took a lot of roadside equipment, like a railroad.

[1] https://www.youtube.com/watch?v=RlZEeIC_2lI

1997 was ten million years ago. Waymo doesn’t require any roadside equipment.

> I'm probably just showing my age, but I like the idea of a "drivered" truck leading driverless trucks versus a completely autonomous system. It's similar to my attitude on crewed spaceflight-- I like the idea of the ingenuity and capacity for independent thought supervising an automated systems, versus autonomous automated systems.

Great example because spacecraft are fully autonomous. The weakest link in that platoon is the driver.

Precious few drunk drivers on orbit. Or activists blocking traffic, or any number of other interesting situations.

Also, it eases the pain of introducing automation by removing a major roadblock to incremental adoption. With the Pareto Principle in mind, get rid of most of the drivers relatively quickly, then work on the holdout cases.

Thise are just obstructions. Call a waymo and experience it for yourself. Humans can’t achieve that level of awareness.

Wouldn’t remote monitoring be better, or the monitor is in the lead truck that is also driverless, monitoring the other trucks in the platoon?

Seriously, we are getting closer to shadowrun rigger territory, but without the neural implants that would have put us deeper in the loop.

That seems likely to cause problems with drivers trying to merge.

Would piracy become a problem in this case? Pirates would only have to deal with one lonely driver to get to multipe trucks’ worth of goods.

The return of the highwayman is exactly what this century needs next.

But in seriousness, what’s the next step? You drive off with five semi trucks worth of stuff? It seems like it’s just a worse way to steal trucks. Especially autonomous trucks that are going to be very well-instrumented and remote accessible.

I mean. https://a-american.com/land-pirates-robbing-from-la-trains/

Completely driverless and I wonder when we’ll read about the first case of truck cargo piracy.

Its probably easier to hijack a truck with a driver, because you can bribe drivers to be the inside guy.

Did you read a lot of classic sci-fi growing up? This sounds very I, Robot.

Reply on news.ycombinator.com