Hacker News new | past | comments | ask | show | jobs | submit login
AI Is About To Take The Ship's Helm Away From Humans (forbes.com)
61 points by howrude 3 months ago | hide | past | web | favorite | 64 comments

Several issues with this idea:

1) On the open ocean, what is there to even avoid? In the harbor, you often don't have a choice; the harbor master is going to _require_ you to pick up a harbor pilot who is rated and certified for the harbor you'll be in and they must be in command of the ships steering and propulsion in order for your ship to even enter their waters.

2) Only 20% of the global merchant fleet is dedicated to shipping. The two other major types of ships, tugs and passenger ships make up most of the rest and neither of these are exactly suited to this technology.

3) Over 50% of the ships in use right now are more than 15 years old. Yea, a new one is _only_ $200 million, but that washes over the real economic picture of ship ownership and maintenance and the likely market for this product.

4) Most ships are not giant Maersk style super-container ships, so staffing reductions aren't likely to be a huge win here, if at all. Ships already run with the _minimum_ crew necessary, usually determined by already established maritime laws and agreements.

5) Microphones and pictures are nice, but ships have actual radar; why not just use that and skip all the fancy "sensor fusion" and off-brand retrofitting?

There are some ships, small sailing ships for example, that do not get picked up by radar. They often need to get radar reflectors fitted to be visible. You certainly wouldn't want to be on a sailing vessel while an automated cargo ship barrels down on you. Sailing ships get right of way to powered ships as well so detecting the difference between vessels is important and the automation would need to take that into account too. Although I am not sure how it actually plays out in practice. I somehow doubt a tanker is giving way to a cruising sail vessel once it is underway.

> Although I am not sure how it actually plays out in practice.

There are exceptions, in established shipping (and a few other types of) channels, the sailboat does _not_ have right of way. Also, when sailboats are using their motors for power they are considered motorboats and the special right of way no longer applies. It is also very rare to see a sailboat under sail power in a harbor and large commercial ships are always manually piloted in a harbor.

(I recently sailed through heavy shipping areas like the channel and Seine on a small sailboat.)

Assuming they can see you, the captain of a large boat will try to give way rather than squash you (they'd be _super_ angry about having to do so, though), but in practice they're already hugely under-crewed, so there's a large percentage of time when you (as a smaller boat) can't count on that. And even if they are actively looking out, small boats are just... hard to see. Almost all of the deaths in sailing come from a small boat's crew falling asleep without appropriate watch (you're meant to fully scan the horizon at least every 20 minutes) and getting run down by a big boat.

This is a long way of saying that big boats are already mostly automated, and I think everyone will be safer if we go all the way on it, along with more robust sensors, instead of just pretending that there's always someone doing a perfect lookout.

> Although I am not sure how it actually plays out in practice.

Radar reflectors are generally legally required on sailboats where it would matter. Offshore most sailboats are outfitted with AIS transceivers and it's legally required in many places for commercial ships.

When not in a port, large ships will often adjust their course slightly to avoid coming too close to a small sailboat even if they're in a shipping channel (if there's reasonable space to do so). With AIS the ship can see exactly where the sailboat is, its heading, and its speed, so it's pretty easy to avoid it. They'll do the same based on radar data if that's all they have. It's common for both boats will tweak their course slightly. Personally, if it's possible, I usually try to adjust my course early enough to avoid a large ship having to react.

I always called it the "right of weight".


Might makes right.

Sailing vessels only get right of way with small vessels. You can't sail your little 20 ft trailer sailer in front of a container ship and expect them to give way.

I can't remember what the actual rule is, but I think it's somewhere around the 250 t mark that sailing boats no longer have right of way.

The exceptions have nothing to do with tonnage and everything to do with maneuverability, either due to the nature of work or the need to stay in a channel to prevent running aground.

Of course you can’t expect a tanker to break/turn on a dime, but there’s nothing in the rules about tonnage.

The Rules of the Road as expressed by RYA instructors are "Give way to big ships". Is it big? If in doubt, yes ...

Completely agree. Quote from the article also kind of shows the founder isn't really correlated numbers properly. "Reading that more than 80% of goods are transported by sea, a light bulb went off, says Vollmer. 'We can have a very huge impact,' he remembers thinking." https://www.statista.com/statistics/264024/number-of-merchan... shows that there are roughly 52,000 container ships in existence. 52,000 container ships moving 80% of the world's goods is something I assume people do not want to be automated and want to keep a very watchful eye on- notwithstanding all of the other issues in executing the idea.

I get the impression you didn't actually read the article, as it addresses most of your questions.

> as it addresses most of your questions.

Vaguely suggesting that three different companies are separately attacking different parts of the problem in a very narrow area of the world with special permits and no collective breakdown of their efforts, their progress or their potential benefits didn't seem to effectively address anything nor does it support it's own headline.

The headline was my main beef; but, we could get into the fact that the two most common maritime insurance claims are for weather damage and theft.. striking submerged objects is also a very common one, which is why harbors require skilled pilots with continual testing to bring boats in and out of those waters.

I'm still very much suspicious of these "bolt-on" AI businesses and all the uncritical hype that surrounds them.

> In the near term, it’s more likely that the helm will be controlled by an autonomous system or a remote human operator while a smaller crew takes care of the vessel.

Hopefully the helm and navigation controls are governed by a secure, air-gapped connection; otherwise someone could hijack the signal and commandeer the ship remotely, and that would be the first case of actual internet piracy.

There has been experiments with GPS successful Spoofing attacks against ships, so even if it is airgapped, it's not 100% safe.


Sure, but then they wouldn't be internet pirates, they would just be electronic pirates.

Air gap is a metaphor dealing with system borders - if you have a fully trusted input, you don't have an air gap, even if that input comes in wirelessly.

“However, autonomy will play out very differently on the water than on land, and in many cases it won’t take humans off the ship—or entirely away from the controls. ...

In the near term, it’s more likely that the helm will be controlled by an autonomous system or a remote human operator while a smaller crew takes care of the vessel.”

Lots of different things rattling around in the ol’ promise of “autonomy.”

I don't foresee this changing anything for ships. All modern vessels, both commercial and for pleasure, have auto pilots and they are used extensively.

The primary job of a captain isn't to drive the vessel, it is to basically be a project manager ensuring that the vessel is running smoothly. This includes everything from crew schedules, maintenance schedules, supplies for the vessel, quality control of procedures, and on and on. The actual piloting is a very small part of what they do.

I could see a potential with completely autonomous vessels that could sail for weeks/months on solar/wind power without any human presence. Crew would only board the ship when approaching a harbour, and steer her to the dock.

Solar panels and wind turbines could generate electricity which would be stored in batteries and then used to drive electric motors propelling the ship. Slower than diesel but also cheaper and with less moving parts.

> Solar panels and wind turbines could generate electricity which would be stored in batteries and then used to drive electric motors propelling the ship.

I wonder what the efficiency of a wind turbine + engine would be compared to a sail.

Ie. if the wind blows so a ship with a sail moves at 20 mph, how fast could the wind turbine + engine move it?

There’s also a lot of maintenance that goes on while the ship is out at sea - requiring crew members regardless of who is piloting the ship. Requiring a ship to go into drydock for routine maintenance is pure loss for the shipping company.

Can someone explain why there aren't that many autonomous _trains_? And I don't mean SFO airport train and others like it, but a real, full-size train moving people or goods around the country. Seems like every other year a train flies off the rails due to nothing more than negligence of train's engineer. Why not automate this completely? You don't even have to steer.

I'm assuming you're talking about the kind of trains pulling 1,000 mile runs as opposed to metro trains.

Train drivers do more than just drive a train. They also have to deal with dropping off and picking up cars from the consist. They also need to monitor track conditions that aren't necessarily easy to pick up from track circuits, such as deep snow (https://www.youtube.com/watch?v=Yja2VmZOfdA shows you how some trains deal with snow).

In general, the costs of upgrading track to be able to support autonomous driving aren't worth the savings you get from eliminating driver. The main counterexception to this is Western Australia, where everything of value is 1,000 miles away from where anybody lives (and is transported by trains to the coasts for export), and so railroads face tremendous costs actually getting their drivers to their trains. Unsurprisingly, the mining companies in Australia who own/run their own railroads have invested heavily in autonomous trains.

The mining operations in Western Australia have also invested heavily in autonomous EVERYTHING.

The massive dump-trucks and drill rigs are going autonomous, with anywhere between 10% and 100% (depending on the site) of the heavy vehicles are now driverless.

Closed environments like mining sites or airports are much more easy to automate.

The routes can be easily predefined, they are relatively short and you can add additional beacons / markers that make things much easiser.

I think the problem with trains outside of say closed loop underground trains that you have a lot of tracks in varying conditions, you can have track abstractions you have a multi agent environment with crossing and that at least as far as cargo trains go the engineer does more than just drive the train they handle many more operations.

The financial incentive for these mining companies are huge though, as a driver can earn A$120k before even considering the total cost of employment for a remote FIFO site.

Mine site are open cast, mine large tonnage and have long mine lives justifying the large initial investment.

Australian mining operations are also very dry, which helps a lot. Rain introduces a lot of variables.

Large scale automation is particular to the Australian mining industry at this point.

South African deep level gold mines have been working on this for many years with little success, and I think they have mostly given up.

The dry conditions help the automated vehicles, but there's special challenges when it DOES actually rain.... everything tends to be underwater, by a lot :-)

not a lot of autonomous trains though... interestingly.

Amusingly, in your video, the view from the driver is completely obstructed by the snow.

From the Wikipedia article on BART (the Bay Area Rapid Transit system):

> Less than a month after the system's opening, on October 2, 1972, an ATC [automated train control] failure caused a train to run off the end of the elevated track at the terminal Fremont station and crash to the ground, injuring four people.

In a bit more detail, they had trackside something-or-others that gave speed information to the train. One train mis-read the speed instruction as calling for 67 MPH instead of 20-something, at a time that it was heading into a dead-end station. When that happens, if there's no human there to say "wait a minute, that's not right", the mistaken command gets obeyed, and the train crashes.

Now: Does that happen more often with automated systems? Or less often? Arguably, less. But to this day, BART has people in the cabs of their trains, at least to monitor the automated system.

The human can look out of the front window and notice that e.g. the cliff above has fallen onto the track, or cows have broken through a fence and are now standing on the track.

Something like the DLR normally operates autonomously, there's a member of staff trained to drive it on every train but it's actually slower and more dangerous if they have to. But that's a completely grade separated system, cows on the DLR is about as likely as cows up the Empire State. And still the first train every morning is manual because maintenance teams are forgetful and a passenger sat at the front doesn't want a metal pry bar in the face when some idiot left it on the track despite checking it off as recovered.

At high speeds a driver can’t do anything if they see an obstruction. Breaking distances on trains tend to be measured in miles, not metres.

Australia uses what are called road trains, which essentially is 3 full sized semi-trailer loads pulled by a massive single cab travelling 90km/h. A good place to find them are the highways between Adelaide and Perth or Adelaide and Darwin. At night, their stopping distance is greater than their headlights can reach; if they see something, it is too late to stop and the choice is swerving and likely tipping or plowing on through. This is why they have bull bars on the front (you might know them as cow catchers), so when there is a herd of cows on the road staring you down the cab can plow straight through without slowing.

(or maybe tech has caught up in the last few decades and they detect obstructions from further away, but that would spoil the story)

Tech would have needed to see through terrain (road is not always flat - even though sometimes the crests and dips are so rare there's a warning sign ahead: https://media.gettyimages.com/videos/outback-road-sign-of-cr... )...and be retrofitted into the existing cabs. So, no: the tool of choice is still brute force, literally.

Sensors would work. If the cab isn't already tall enough (they are big!), you could mount the sensors on a pole. The road isn't always flat, but flat enough; I doubt a road train can deal with much of a gradient.

They could carry a set of battery packs on the roof and a pair of drones acting as scouts, one flying ahead with cameras and lidar for 10 minutes while the other one returns to swap out its battery.

Cool...but you need three to have backup, and a battery swap mechanism, and a charging system, and this contraption happens to be more complex than the rest of the car combined. I mean, it could work...but it feels very Inspector Gadget-ish.

What happens if there's a broken car/truck blocking the road?

The same thing that happens when any road vehicle meets a break down they didn't see in time, except the massive truck is more likely to win.

http://divine.vic.gov.au/lifestyle/health-and-wellbeing/road... has an account of a collision with a road train:

"The road train driver had slowed down to make a left-hand turn at a crossroads and Scotty had slammed into him at over 100 kilometres an hour. According to the autopsy, Scotty was over the legal alcohol limit. The driver of the road train kept going once he had completed the corner. I was later told he thought he felt a shift in the load as he slowed the vehicle to negotiate the corner."

Anyone wanting to make autonomous trucks perfect should remember the bar is actually pretty low...

Braking to a stop is the best option (which is not always available, yes). Emergency braking to half the speed is the difference between "8 fatalities" and "hundreds of fatalities"; Ek=mv^2 - note that velocity squared*.

Case in point: https://en.wikipedia.org/wiki/2008_Stud%C3%A9nka_train_wreck - train hits bridge at 90 kph instead of 135 kph; a 2/3 speed, but less than 1/2 the kinetic energy.

They can, however, hit the brakes as best they can (kinetic energy is proportional to the square of velocity, so this helps a lot), and warn passengers to brace for impact (example: https://en.m.wikipedia.org/wiki/Glenbrook_rail_accident#Effe...)

I agree drivers have other benefits. I will say, you made me realize that an electric sled with 3D LIDAR could check our track hundreds or thousands of meters in front of the train and confirm the track was still safe.

What? There are dozens of installations worldwide with many more on the way.


If you are talking about intercity trains instead of metro systems - although metros typically handle many times the number of passengers - China is currently trialling driverless technologies on its Fuxing class HSR on the Beijing-Zhangjiakou route with expected entry into service by 2022.

Sure, there are autonomous trains in some places, I said so right in my post. But why aren't _all_ of them autonomous by now?

The technology's only really been perfected in the last 10 or 20 years, well below the normal replacement cycle for this kind of heavy infrastructure. Give it time - it's inevitable.

Because railway lines, in the past, were not designed with autonomous operation in mind. It’s really only suitable for completely isolated systems (high speed bullet-train style lines walled off from their surroundings, or Western Australian iron ore trains miles away from anyone) or lightweight metro/subway style trains (e.g. DLR, airport “air trains”) that travel at low speeds or can stop quickly.

Upgrading is hard, retrofitting is harder (e.g. NYC subway runs on a century-old technology), and most importantly, it all comes with a hefty price tag, all for an incremental improvement only. Follow the money.

Dockland light railway is completely automated. This train line is in London. I have travelled on it. There is no engine room or train driver. It felt interesting to sit in the very first train cabin which would usually be an engine room in other trains.

Except when the guard takes over driving, which happens a lot

It's in the works: https://en.wikipedia.org/wiki/Automatic_train_operation

I think unions will get in the way as much as they can, unfortunately. As an example, a 'fireman' rode along on diesel and oil trains LONG after they were rendered obsolete by the fact that there was no longer any coal to shovel.

We've had them in Australia's north west for a long time. Interestingly they come with their own set of problems.


> "...a driverless train..."

> "The driver had stepped off the train..."

It wasn't a driverless train (except in the technical sense that the driver wasn't onboard) and the quote comes from CFMEU (for non-locals, that's the Construction, Forestry, Mining and Energy Union, a massive workers' union) which has a vested interest in pushing back against automation.

The 'driver' operates a dead man switch in a lot of cases, and that's about it. I can't recall if this train itself was an automated test bed, but a lot of BHP and Rio's rail operations are managed from the control center in Perth, and drivers are (usually :P ) aboard in case of problems.


So it looks like even though they've had the technology going for a while, Rio only just delivered their first 'real' loaded autonomous ore train last year : https://www.riotinto.com/ourcommitment/spotlight-18130_25692...

More on the AutoHaul system: http://www.calibregroup.com/projects/project/autohaul-projec...

Apologies for the mixup

Easy mistake to make, the story was a bit misleading calling it 'driverless'!

http://www.calibregroup.com/projects/project/autohaul-projec... is the physically largest autonomous system in the world.

I wonder about that too. It seems a much simpler and constrained task than driving a car.

> It seems a much simpler and constrained task than driving a car.

It's not. Always assume that whatever limit can be exploited to maximize economic return is pursued to the maximum extent by everyone involved at nearly all times. If you do so you will be right more often than you are wrong. An occupation that is conceptually "easy" is made demanding through optimization.

Heavy freight trains that navigate hilly or mountainous terrain can rip themselves apart when they are not operated with precision. Passenger train operators can be held to schedules where deviation in either direction is measured in seconds regardless of weather, faults, mistakes, uncooperative passengers, etc.

When large amounts of capital are involved always assume that the expectations will be extraordinary and, therefore, difficult to automate.

I wonder if that's on purpose, but I interpreted the title as metaphorical first, thinking it was another dystopian fantasy about computers taking over the world.

I went on a similar, but slightly different path. I thought it was an article not so much about a dystopian fantasy, but about the harsh (and in some ways also dystopian) reality of AI and ML systems not being "debuggable" and understandable by humans like traditional algorithmic systems had been, which effectively means that we can only reason about decisions of such systems to a very limited degree.

No, not just you, exactly how I read it too

I thought so but then I saw that it was forbes and quickly figured it was simply a pun and it probably refers to some kind of automated sailing system.

I've only read the first paragraph of TFA and it looks like I was right.

Remember the movie Hackers when the DiVinci virus capsized those ships?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact