Autonomous Mayflower to attempt Atlantic crossing, again
- Reference: 1651212134
- News link: https://www.theregister.co.uk/2022/04/29/ibm_autonomous_mayflower/
- Source link:
Named after the 17th century vessel that carried over 100 passengers from England to what is now Massachusetts, the MAS was [1]initially designed to retrace the historic route – [2]but failed at the first attempt. This time, it'll be going from Plymouth, UK to the US capital city Washington DC instead.
The MAS began its 3,000-odd mile journey yesterday. Armed with four computers containing Intel processors, six Nvidia Jetson AGX Xavier GPUs, two Nvidia Jetson Xavier NX boards, and a few other custom chips, the ship processes incoming data from a suite of cameras and sensors with machine learning software. Computer vision algorithms and radar help identify things like shipping containers, cargo ships, or fishing vessels to navigate around obstacles.
[3]
If any of its hardware breaks down, however, there's no one around to fix it – which proved to be a problem last time. Modern large ships generally have a human-operated machine shop and a cache of spare parts to install for just these sorts of eventualities.
[4]
[5]
Still, Promare, a maritime non-profit, and IBM, which helped build the 15-meter autonomous sailboat, are confident it can make the journey by itself. The craft uses GPS to keep track of its path, and takes into account external factors like weather conditions as it motors along.
The MAS relies mostly on its solar-powered engines for propulsion, but has a backup diesel generator just in case. Last time, the ship broke down due to a [6]mechanical issue in that generator that borked the fossil-fuel engine. With only solar power, the ship started slowing down and would eventually have run out of energy.
[7]
Brett Phaneuf, co-director of the project and a previous board member and president at ProMare, told The Register the team had improved the computer vision software and its mechanical systems to make the boat more robust since the failure. The ship was cleaned up, and old kit was replaced to get it back in working condition.
[8]Self-sailing Mayflower ship to have another crack at Atlantic crossing next year
[9]Mayflower, the AI ship sent to sail from the UK to the US with no humans, made it three days before breaking down
[10]The human-devoid AI-powered Saildrone Surveyor ship just made it to Hawaii from SF
[11]UK funds hydrogen-powered cargo submarine to torpedo maritime emissions by 2050
The MAS project is designed to see how autonomous seafaring technology can support ocean and climate research in the future. As the ship motors to the New World, it'll also be taking water samples to study microplastics and algae and listening out for whales and other marine mammals.
"The [Atlantic] crossing, from our point of view, is not the ultimate goal. But once the vehicle can prove that it can run without problems for this length of time, we can then plan on using it for further science expeditions in remote areas," Ayse Atauz Phaneuf, Promare's current president, told us. "The real gain in all this has been to develop an integrate this technology, and test it on Mayflower."
But before they can think about their next journey, the ProMare team is hoping they'll successfully complete this trip the second time round. "We are confident we'll get a long way across and that we'll make it – but the ocean is the biggest challenge now as there are few other obstacles to overcome once we're out in the deep blue. For the moment the ocean is cooperating and we hope that it continues to do so - fingers crossed," Phaneuf said.
You can watch the ship via a live stream posted [12]here . ®
Get our [13]Tech Resources
[1] https://www.theregister.com/2021/06/17/ibm_autonomous_ship/
[2] https://www.theregister.com/2021/06/19/in_brief_ai/
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2Ymu3TKLe6hY2J6SnAnjPUQAAAFE&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Ymu3TKLe6hY2J6SnAnjPUQAAAFE&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Ymu3TKLe6hY2J6SnAnjPUQAAAFE&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://www.theregister.com/2021/09/27/autonomous_mayflower_ship/
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Ymu3TKLe6hY2J6SnAnjPUQAAAFE&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[8] https://www.theregister.com/2021/09/27/autonomous_mayflower_ship/
[9] https://www.theregister.com/2021/06/19/in_brief_ai/
[10] https://www.theregister.com/2021/07/09/saildrone_surveyor_radar/
[11] https://www.theregister.com/2021/09/16/hydrogen_cargo_sub/
[12] https://mas400.com/
[13] https://whitepapers.theregister.com/
Re: Not the ultimate goal ?
There are probably some good uses.
A good example would be searching for wreckage on the ocean floor.
When the Air Malaysia flight disappeared off Western Australia a few years ago, ships spent days/weeks scanning the ocean floor.
If you could get a fleet of these automated ones to head out, scan an area and report back, I be think that would be pretty useful.
Ideally they'd run off solar/wind and not rely on fossil fuel (removing the need to return for refuel).
The fact that they're still trying to just get the ship to sail in a straight line between two points suggests we may be still quite a way off my example though - baby steps ;)
Re: Not the ultimate goal ?
Honestly, an unmanned ship is good for what if it is not transporting cargo ? It's not like you can task it with conducting scientific surveys on land.
I suggest you go and speak to NASA...
Re: Not the ultimate goal ?
So what is it you actually want this for, taking pics of penguins?
It's a proof-of-concept. The eventual aim is to sell the technology to P&O Ferries who then won't have to pay anything to their ships' crews...
Re: Not the ultimate goal ?
the only personnel you might be able to eliminate on a cargo vessel is the captain and maybe two or three people directly below him like the helmsman and first mate. The rest of a ships crew are (as the article already indicated) always busy with maintenance tasks to keep the vessel in good condition. Doing that while underway means far less time spent in port (which costs money) and far more time underway (which means less time between payouts). And even the captain isn't "useless".
out of the many classes of ship, the big cargo ships are the LEAST likely to get automated any time soon imho.
Fair winds
I hope it makes it this time though the idea of a 300000 tonne oil tanker with no humans is a scary prospect .
Icon because - Arrh Jim me lad!
Re: Fair winds
Perhaps that's not as bad as you are thinking. Tugboats manage those big ships around ports, so the "self driving" option would only be in open ocean. Even then, a ship keeping lookout via radar, lidar and sonar is probably going to do a better job seeing some tiny craft half hidden by waves than a bored and possibly tired and/or drunk captain.
Much much less to hit in the ocean than on any road other than maybe those in the Outback.
Re: Fair winds
It's not the risk of collision to worry about, more the opportunities for piracy. That 300000 tonnes of oil could fetch quite a tidy sum!
(or more likely, just a few tonnes of it at a time - helps with both the logistics and the risk of being caught)
Not the ultimate goal ?
So what is it you actually want this for, taking pics of penguins ?
Honestly, an unmanned ship is good for what if it is not transporting cargo ? It's not like you can task it with conducting scientific surveys on land.
Sure, you could tell it to go to a given "remote point", and then what ? It takes some samples of seawater and a video of the shore ? And ?
The only reason for automating a ship is to make cargo ships that don't need personnel on board. Anything else is just pie-in-the-sky wishful thinking.