IBM's autonomous Mayflower ship breaks down in second transatlantic attempt The crewless AI-powered Mayflower ship, which was on its second attempt to cross the Atlantic ocean alone, is being hauled back to shore after suffering a mechanical failure. The ship, built by ProMare, a non-profit organization focused on marine research, with help from IBM, set off from Plymouth, UK, last month with the goal …
"...they should try smaller scale tests before heading out to the Atlantic again. Possibly a trip across the Serpentine. Then incremental stages from there."
Here's someone who could probably give them a few pointers - and get them past the first few increments in one step.
https://www.youtube.com/watch?v=90N8w69Sqlk
https://www.youtube.com/watch?v=HFhsh4V8SZA
I came to the same conclusion about the next bit of We Sailed the Ocean Blue. It starts with balls flying. Did people really express themselves that differently then or did Gilbert get away with a massive double entendre? I suspect some of the gentlemen in the audience had a snigger.
Large Balls of Copper were hoisted as signals. Still used in 1970s at Leith. Dropping of ball indicated exactly (within a few seconds of 1pm at same time as one o'clock gun was fired from Edinburgh Castle to give a time signal to mariners in Leith Roads.
Refer
https://stories.miln.eu/graham/2014-01-09-one-oclock-gun-at-edinburgh-castle/2013-edinburgh-one-o-clock-gun.jpg
"The glitch appears to stem from the ship's generator
Ho many articles have we seen about data centers brought low by generator problems.
They're not space boffins who deserve a pint, just computer nerds who are still working on it so so I'd offer some strong coffee to help while they fix it and wish them more luck on their next attempt.
At least the Azores has a much better climate than Plymouth (UK or US).
It is all very well making it autonomous and ai driven, but it seems to me the problem is in the engineering of the boat. I mean nasa can sling a rover to mars land it and then keep on running way beyond the design life, so far this has managed to keep going for two weeks.
The IBM website makes great noise about the 30 ai sensors it uses to navigate. How much of the processing is devoted to keeping its systems running.
> The IBM website makes great noise about the 30 ai sensors it uses to navigate.
And that's the problem: I guess they assumed the rest would "just work".
A ship, especially such a small one, is a very different and extremely challenging environment (extremely humid, constantly and violently rocking and rolling), you can't assume it's just a mobile data center and some waterproofing is all it takes. You need a firm of (competent) shipbuilders, not a bunch of IT guys, especially when expecting to sail across the Atlantic (which is rather biggish, I know, I've done it).
During my journey towards becoming an old duffer, I have noticed a distinctive shift in management style.
A new type of leadership is rising, men in their 30's, who are very good at talking, presenting and selling, while being even better at never implementing any of it, so they can sell more, faster, get more "wins", and move up faster, than we BOFs who are hindered by also implementing what we just sold.
Do you suppose somebody should tell them that Harbor Freight's chineseium gensets aren't built for continuous duty, especially in a marine environment? Talk to the folks who ship things that need refrigeration via container ship. The gensets that power the refer units are built to go months without human supervision. I'd recommend starting with SeaBox or the like ... Most such companies do custom kit, and I'm sure they'd welcome the advertising. Might even do a free installation (with redundancy) in return for prominent largish sticker placement.
Besides, this really isn't something new. Autopilot for civilian boats has been around for over half a century. For example, the unit I pulled out of my Monterey Clipper was a 1970's system based on LORAN C and made by Raytheon. It has since been replaced by something that'll get me a trifle further off the coast ... but I only use it to take me from harbo(u)r to harbo(u)r in SF bay and up and down the California coast. Overkill? Absolutely ... but it was cheap. My buddy can (and has!) get into his boat and program a route to Hawai'i. It'll happily take him there (and back) without any supervision, including avoiding storms and other shipping, while the occupants of the boat party
One does rather feel that there are generations of ship engineers who know how to build things that can survive, say, waves... perhaps the designers might have consulted them at some stage? I can't help feeling that they've gone all excited over spiffy new software and kind of ignored the basics.
As someone else pointed out, it's not as if this floating on water is new technology... maybe they should have just bolted their automation onto a crusty old fishing trawler and started from there?
Either this is a pet C-Suite project or someone has managed to convince the C-Suite this is a money making idea for IBM, regardless of any pre-existing and proven method for auto-navigating the seas.
IBM don't do anything unless there is a perception they can screw people out of money with overpriced and overly complicated crap.
I can't even imagine what would happen if it broke down. What would IBM do? Send out an offshore engineer to come fix it?
Navigation is the easy part really. To keep the boat afloat and running is the hard bit.
Generator , engines , pumps , hull .. there's a variety of problems and challenges every trip.
Good preparation and maintenance is one , but since anything that can go wrong will go wrong at sea it seems evident there's need to keep a crew on every ship. Well .. if you want the ship to reach destination that is. If you don't care .. that's another matter.
I read the dimensions and of course, immediately thought that it has to be a massively scaled down model. No way would 130 people trundle across the Atlantic in a 15m boat.
Then I read that the real one was merely twice the length. Even accounting for multiple decks that just blows my mind. How they didn't all murder each other...
"How they didn't all murder each other..."
Ever been in a small boat in the North Atlantic from mid-September to mid-November?
Most of them were probably all lined up along the rail praying ... "Oh, god" ::hork:: "Oh, god: ::hoorrk:: "Oh, god" HOOOORRRRK!:: etc. etc. The few who didn't get seasick were kept busy trying to keep the rest hydrated. No time or energy for fisticuffs under those conditions.
I rather suspect that the genetics of the 47 survivors are somewhat diluted by now.
On the other hand, the Puritan nutters (which we were quite happy to throw back at Blighty as soon as possible) went on to take over the UK almost entirely ... You can still quite clearly see the affects all these years later.
It was a civilian ship in the early 1600s. I seriously doubt they were using anything resembling modern watchkeeping methods. Only the bare-bones number necessary would have been above decks most of the time, the rest would have been huddled below (well, in the poop deck) trying to get dry and cold (as opposed to near frozen ... warm would have been out of the question, unless you were the cook, the Master[0] or a mate).
The passengers (102 people), not being seafarers, had nothing to do with the running of the ship. That's why they hired ~30 crew. It's also why they got suckered into hiring a ship that had been scrapped a couple years previously... and one that had been built for the relative ease and comfort of the cross-channel trade. There is a huge difference between only one to three days port to port (London to Bordeaux & back, maybe a side trip to Amsterdam occasionally, certainly never as far as Stockholm or Helsinki or St. Petersburg.) and open ocean work with months between ports. It was also built for bulk cargo, not passenger comfort.
Thirdly, read the history of the trip. The weather wasn't exactly conducive for sunbathing. In fact, it was so rough that one crewman was washed overboard and lost, and a passenger who was also washed overboard managed to grab a rope and was rescued. All accounts state they had difficulties keeping dry, even below decks.
Fun trip! It's little wonder that the after affects, combined with a nice New England winter that they were ill prepared for, managed to kill off over half of the voyage's survivors ...
[0] "Captain" was a military rank, and not used in this civilian context back then.
Replace IBM with Nice Automatic Ship Again ...
These problems that we're seeing probably indicate that the design team built the ship as "working" but never verified that if wouldn't fail. That's pretty much a normal design environment these days but the NASA engineers are totally devoted to making their designs work for years.
Upvoted, but to be honest it has nothing to do with crossing anything since apparently they always fail because of the same basic shipbuilding errors. Clearly they've spent all their efforts and money on the IT part, and just rushed the parts which actually were way more important for the success of their project.
You need to keep in mind the "AI sails the Atlantic" event is just a publicity stunt. As Jake said further up, GPS-driven autopilots are common as dirt, and if coupled with AIS they even can avoid other ships (which in the middle of the Atlantic are extremely rare anyway). And surprise, those reliable autopilots don't need any "AI" (or blockchain or whatever) at all! IBM is just re-inventing the wheel, adding some "AI" to justify their presence...
There is a massive amount of interest and discussion about fully autonomous commercial shipping. Don't forget that most of the world's trade still travels on ships. and they take a relatively long time to cross the oceans. The shipping companies want to get rid of crews but there is (rightly) big concern over the legal and ethical issues (maritime law has been around a long time). As well as practical issues of how loss rates would change (would there be more accidents and losses, or fewer?).
It is a shame that this project seems to have been let down by the mechanical engineering. And embarrassing for IBM to have their name associated with i!
This has already been discussed in the comments of the previous article about this.
Crews aren't there to navigate (autopilot and all), they are there to keep the vessel in working condition. Unlike a car which only needs an oil change once a year, a ship navigating in a hostile and extremely corrosive environment needs constant (as in "every day") care. Things break down, start rusting, and there is no way you will stop a container ship for repairs, she costs you many hundreds of thousands every day no matter if she's earning you money or just sitting around. So repairs are done while underway. The only acceptable shore time are the 10-20 hours while she's unloaded and reloaded, and that's an all hands on deck affair.
A crewless container ship wouldn't be an economy, on the contrary she would be a pending financial disaster (sending one of the big ocean going tugs out to capture her and bring her in for repairs would cost you more than paying a crew would ever have). Ships are not just oversized seagoing trucks!
Its not clear the industry agrees with you... https://www.imo.org/en/MediaCentre/HotTopics/Pages/Autonomous-shipping.aspx
Personally, I think the main real goal is to get rid of expensive people like Masters and just have a couple of maintenance people on board recruited for almost no money in low-wage countries.
"I think the main real goal is to get rid of expensive people like Masters"
The cost of a Master, per year, is far, far lower than the cost of (for example) replacing a propeller or rudder on a container ship. The downtime alone is horrendous, never mind the hardware.
And if the shaft is bent, and the stuffing box is damaged ...
> Its not clear the industry agrees with you...
To quote your article: "IMO wants to ensure that the regulatory framework [...] keeps pace with technological developments" (emphasis mine). The IMO is a UN regulatory agency, not shipowners, I guess they just regulate ahead just in case.
In reality the crew is a vanishingly small part of what a ship actually costs a day (mainly fuel, depreciation, insurance/fees/taxes, maintenance).
On the other hand a well trained and experienced crew is a guarantee that your ship will keep deadlines and will have no expensive downtime or, worse, accident. Remember the Amoco Cadiz? Her rudder jammed, and as a result she went down, ship and her cargo were lost, and the company had to pay $230 million in fines and compensations for the oil spill. Now that's a shipowner's nightmare, not the modest wages he pays some Filipino crew (not derogatory, the Philippines train the best crews - and they are cheap too!).
As for the captain, talking about getting rid of him is like talking about replacing a company's CEO with an app! They might have the biggest salaries, but they are essential for running the place (navigation is just a tiny part of their job description) and can't be simply replaced by some computer program, no matter how spiked with AI (blockchain, cloud, NFT).
I have no opinion on whether it is practical or even desirable.
I just pointed out that "There is a massive amount of interest and discussion about fully autonomous commercial shipping", which is undeniable (that IMO URL was in a section called "Hot Topics", after all).
And when that level of interest happens, there are always people circulating to make money off the topic. I assume that is what IBM wanted to do (as well as conduct some interesting research). It is a shame that they have been let down by the boring mechanical engineering and not had a chance to do the interesting parts of the research.
Never mind the buzz, it's just hot air convection.
As for IBM, this is clearly a publicity stunt, since there is nothing new or revolutionary in this ship, except the lack of a meatbag on board.
With the necessary funding anybody could build a similar boat and send her across the Atlantic. All you need is an off-the-shelf GPS-linked autopilot with AIS integration. And to wait for a favorable meteorological window to start your crossing. Heck, you could even make your ship use sail power, the automation to furl/unfurl sails and set them according to wind already exists.
No programmer needed, just an electrician to connect the cables... So, what's the added value of IBM here?
And they are safely docked ... I just watched a couple blokes[0] fiddle about with the bowlines. Almost looked like they knew what they were doing.
Beers all 'round for this phase being completed.
[0] Was one a blokette? Hard to tell, poor quality video. Thanks, IBM. So much for your supposed "technology showcase".
This is the problem with autonomous shipping - you can have the most advanced AI ever made, but it'll never be able to wield a spanner or give something a good whack with a hammer when something goes physically wrong. No, for that you would need a robot repair crew, and another robot repair crew to repair the robot repair crew when that went wrong. Then there'll soon be the ships full of robot pirates looking to commandeer your craft mid-ocean. Not to mention the landlubber hacker crews trying to take control. All in all, it's a recipe for lots of expensive hardware to end up in Davy Jones Locker.
Indeed. If you want to cross the Atlantic with a yacht-sized boat, the one and only way is to reach the Azores and then head west. Going straight through the north Atlantic works for big ships, but would be suicide for such a small 15 m boat.
A cousin of mine sailed the Channel->Azores->Caribbean route just a month ago with a 45 feet ketch, and it was a bumpy ride. The bay of Biscay (Brittany to northern Spain) is reputed to be a nasty place, and according to my cousin it didn't fail its reputation. Given the Mayflower is just one meter bigger than his boat, I guess she was tossed just the same, which would explain why the IBM-installed generator decided to throw the towel.
A mate did a few crossings of the Atlantic as an engineer on banana boats. Apparenty stuff breaks all the time and they had to fix it at sea in their workshop. They could radio for help if things got really serious but in practice that was only ever going to happen if someone life depended on it, in all other cases they made it up as they went along.
I'm happy to believe a computer can manage the basics of navigation and watching all the critical parts of a ship all on it's own, however without some pretty sophisticated robotics it's going to be hardpressed to fix much...
Oh, let's save some money by booting the drivers. Yeah, then all of us execs can get big fat bonuses.
Ever seen looters pillage a trailer as soon as it's comes nearly to a stop? I've seen videos of people breaking into trucks on the move. A driverless truck only takes over the much more efficient long haul shipping that a train would do. At either end there are documents to process, lading bills to sign, queues to navigate at shipyards and train depots... and Walmart distribution warehouses. Will an automated truck detect that it's pushing a small car whose idiot driver managed to get glued to the bumper? What happens when rain/snow/mud/insects obscure a sensor to the point where it isn't working correctly, but not completely incapacitated? Tesla automation already shows a propensity for having a blind spot when it comes to emergency vehicles. What happens when a patrol car is blocking a road because the bridge ahead is out and there hasn't been time to get a proper detour set up?
I suppose that the ocean is much bigger with far less to run into, but it's not completely without things to bash around. Is there any information on the average retail value of the contents of a large container ship. At least well into the billions. I figure it would be good insurance to have a crew onboard to be able to turn everything off and then on again once in a while if the ship stops working. Even if they couldn't effect repairs, they could inform the home office what they need to get the ship going again and it could be sent to them or they could get a tow.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
