Automated Ships?

[Spirit of Glenans]

I've just spotted, on the right of my screen, a news item concerning a project to build fully automated, crewless ships, to initially be remotely operated, but eventually to be made fully autonomous, i.e. no human intervention.
I wonder will their watchkeeping efficacy be better or worse than is currently displayed by many commercial vessels (allegedly )?

 

[westhinder]

I am very curious how the operator, or at a later stage the ship itself will distinguish between a motor boat and a sailing boat at night on a bouncy sea and how it will deal with colregs

 

[Kelpie]

Given the advances being made in driverless vehicles, my doubts are not based around watchkeeping or COLREGs where I reckon a roboship would be far superior to any coaster manned by a half drunk sleep deprived human.

It's the practical aspects that I wonder about- how is it going to tie up? Are they going to have little catapults to throw the warps ashore? Are they going to have a robot engineer with big robot hammer to dole out percussive maintenance as required?

 

[Kukri]

Given the advances being made in driverless vehicles, my doubts are not based around watchkeeping or COLREGs where I reckon a roboship would be far superior to any coaster manned by a half drunk sleep deprived human.

It's the practical aspects that I wonder about- how is it going to tie up? Are they going to have little catapults to throw the warps ashore? Are they going to have a robot engineer with big robot hammer to dole out percussive maintenance as required?

I run a fleet of large containerships. The idea of an unmanned ship has been around since the 1980's, and indeed large ships have been built to be run by a crew of 6....

... and scrapped, long ago. The industry has settled on 16-18 as "the right number" and has nil. zero, zilch interest in unmanned ships. You need warm bodies to tie the thing up and once you have put them on board to do that they may as well stay on board, keep a look out, and do routine maintenance.

 

[maxi77]

I am very curious how the operator, or at a later stage the ship itself will distinguish between a motor boat and a sailing boat at night on a bouncy sea and how it will deal with colregs

As the main way of a human deciding what the night target is is by looking at the lights a machine can do this very easily and probably more reliability than many human operators. Colregs are based on logic andare well within the capacity of a machine to deal with. The big potential problems are in the area of sensor failure, but I would suspect most machine sensors are more reliable than human ones. Having before retirement worked in automatic image recognition in demanding environments, much if not all of the technology and techniques are already in being it is more a matter of integrating the system into a ship. It certainly seems to be the way the military are moving

 

[PhillM]

Last I heard the idea was that the automated ship would be met s few miles from port with a pilot and crew who would handle the "close to the shore" stuff.

 

[Kelpie]

Last I heard the idea was that the automated ship would be met s few miles from port with a pilot and crew who would handle the "close to the shore" stuff.

That addresses the problems I was talking about, thanks. Presumably you'd need to transfer a fair number of bodies if they're going to do all the ropework. Is pilot transfer recognised as a high risk activity? Certainly looks, uh, fun...

 

[PuffTheMagicDragon]

It certainly seems to be the way the military are moving

I don't think that the military loses much sleep about what they call, rather euphemistically, 'Collateral Damage'. Perhaps instead of "friendly fire" we shall now be hearing about "friendly collisions"...

 

[Neil_Y]

This one is out there running around, with crew at first but the drivers pod at the front will be removed so it looks like this... keep your eyes peeled..

 

[maxi77]

I don't think that the military loses much sleep about what they call, rather euphemistically, 'Collateral Damage'. Perhaps instead of "friendly fire" we shall now be hearing about "friendly collisions"...

With modern electronics and software system failure is much rarer than human error, so called friendly incidents are always human error. Equally electronics do not get tired, human look outs start losing effectiveness after about 20 minutes

 

[Kukri]

Last I heard the idea was that the automated ship would be met s few miles from port with a pilot and crew who would handle the "close to the shore" stuff.

Somebody hasn't costed that. Suppose the ship has to anchor and wait for a berth - a very common situation... and suppose three ships arrive and two are sailing on the same tide..

Pilot transfer is a high risk, high cost, activity. If we are going to transfer enough bodies to moor up, the risks are multiplied, and one dead body costs a lot, now.

It's cheaper to keep the warm bodies on the ship and employ them on maintenance, cargo care and keeping a lookout on the sea legs.

Hard though it is to believe, merchant shipping is not run by idiots; it is one of the biggest users of IT on the planet. If it made economic sense to have unmanned ships, we could have had them thirty years ago. It doesn't make sense.

 

[Kelpie]

It seems plausible that an unmanned ship could berth itself in a specially designed facility (e.g. it could power on to a well fendered dock) but that would restrict its usage to certain routes only. And the cost of building the dock would not be insignificant.

 

[JumbleDuck]

As the main way of a human deciding what the night target is is by looking at the lights a machine can do this very easily and probably more reliability than many human operators. Colregs are based on logic andare well within the capacity of a machine to deal with.

How would the machines deal with the many, many gaps and contradictions within IRPCS? When A is required to give way to B but B is required not to impede A, for example? Or when one vessel is simultaneously give-way and stand-on to two others? Perhaps that's where the remote monitoring would come in, to do the Rule 2 stuff when the others don't help.

 

[Neil_Y]

The ACTUV vessel is designed to operate with minimal input ( so not completely autonomous ) but I guess it will be able to ask for assistance from a human (remote) operator if it doesn't understand the situation. So you could have a few human operators at shore bases able to dial into a vessel and see sensor outputs and camera views, much like the drones. That way one operator could easily monitor several vessels.

 

[westhinder]

The ACTUV vessel is designed to operate with minimal input ( so not completely autonomous ) but I guess it will be able to ask for assistance from a human (remote) operator if it doesn't understand the situation. So you could have a few human operators at shore bases able to dial into a vessel and see sensor outputs and camera views, much like the drones. That way one operator could easily monitor several vessels.

These operators might be quite busy at times, judging by the amount of radio traffic there is on the pilots' channels off the Belgian coast (my home waters). I do not see autonomous unmanned ships negotiating the approach to Zeebrugge or the Westerschelde, nor for that matter, the Thames Estuary or the approaches to Felixtowe or Southampton.
The prospect might seem attractive for long monotonous and uneventfull ocean passages, but if they have to meet up with a crew somewhere off the harbour to nurse the ship into its berth, I can see all kinds of problems developing. Minn's views are the realistic approach.
All this does not mean that the research is not worthwhile, as there will be scenarios where unmanned vessels can be an option, and at the very least the research is likely to develop better systems for controlling ships and better assistance for watchkeepers. Which will be to the benefit of all who go to sea.

 

[JohnGC]

How would the machines deal with the many, many gaps and contradictions within IRPCS? When A is required to give way to B but B is required not to impede A, for example? Or when one vessel is simultaneously give-way and stand-on to two others? Perhaps that's where the remote monitoring would come in, to do the Rule 2 stuff when the others don't help.

I think the enabling technology will be synthetic aperture satellite radar with wide bandwidth down links to the vessels and their land stations. Much of the hardware is already in orbit and more is planned. It is already being used with AIS and VMS data to police illegal fishing, the coverage is global. The UK has a number of satellite applications development centres and autonomous marine navigation is being discussed.

Together with vessel based GPS receivers, plotters, RADAR and AIS, a lot of "situational awareness" data can be made available without recourse to visible light.

Algorithms will still need developing which reliably and safely apply the data to the COLREGS in real time. Rather like the work being done by Google and others on autonomous cars. Although clearly there is still some way to go for those. I suspect there would need to be changes to the COLREGS to allow for these vessels.

We also have drone technology to draw on. Fully autonomous vessels in the open ocean and land based drone like bridges when the vessels are closer to shore.

Currently the merchant shipping fleets have a lot of life left in them and they where not designed for autonomous navigation. Until there is a good financial reason to change the industry is unlikely to. But in the mean time the technology is being developed.

 

[alanch]

.

Pilot transfer is a high risk, high cost, activity. If we are going to transfer enough bodies to moor up, the risks are multiplied, and one dead body costs a lot, now.

I draw your attention to the death of a Pilot boarding a vessel at Gravesend last month.

 

[Kukri]

I draw your attention to the death of a Pilot boarding a vessel at Gravesend last month.

Yes, that one was a shocker, given the location.

 

[Deleted member 36384]

Automated ships will come along soon enough, it is almost inevitable. Today we have dynamic positioning that is very accurate but can also adjust the level of accuracy depending on the required finesse needed (which also affects rate of fuel consumption). There are probably a couple of thousand DP vessels bobbing about the world these days that go to where they are told and stay within a meter or so of the target spot. You think it is not safe? Many of the DP positioned vessels are bobbing about right up against very hazardous installations. There is no reason why a DP vessel could not steer itself into a berth and hold it's position while being unloaded, securing with warps is not required for safe, secure berthing. The level of DP control and redundancy require to provide assurance is already understood with the technology in place.

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

The IRPCs will be changed to accommodate unmanned vessels, perhaps with an exclusion zone around them as well as electronic, visual and audible identification features. At the moment a DP drill ship has an exclusion zone around it when in DP mode. I doubt that IRPCs will be a hurdle when it comes to implementing unmanned ships and probably the IMO already has an opinion on it.

As said by others the expense of DP is not worth the economics compared to simple technology on manned vessels. Indeed with low cost labour, reliable low cost legacy technologies why bother? There is a reason a DP deep water drill ship costs $750k a day to run in drilling mode.

 

[maxi77]

Automated ships will come along soon enough, it is almost inevitable. Today we have dynamic positioning that is very accurate but can also adjust the level of accuracy depending on the required finesse needed (which also affects rate of fuel consumption). There are probably a couple of thousand DP vessels bobbing about the world these days that go to where they are told and stay within a meter or so of the target spot. You think it is not safe? Many of the DP positioned vessels are bobbing about right up against very hazardous installations. There is no reason why a DP vessel could not steer itself into a berth and hold it's position while being unloaded, securing with warps is not required for safe, secure berthing. The level of DP control and redundancy require to provide assurance is already understood with the technology in place.

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

The IRPCs will be changed to accommodate unmanned vessels, perhaps with an exclusion zone around them as well as electronic, visual and audible identification features. At the moment a DP drill ship has an exclusion zone around it when in DP mode. I doubt that IRPCs will be a hurdle when it comes to implementing unmanned ships and probably the IMO already has an opinion on it.

As said by others the expense of DP is not worth the economics compared to simple technology on manned vessels. Indeed with low cost labour, reliable low cost legacy technologies why bother? There is a reason a DP deep water drill ship costs $750k a day to run in drilling mode.

I agree it is a cost problem, no cash savings no automation. However much of the technology exists and we understand the problems of integration of the variety of systems needed to achieve a solution. The securing alongside problem is less demanding in terms of machinery than DP rigs and should be achievable with not too much extra hardware. We can dock space ships on auto. As for IRPCS in general they are a logic problem and a machine is often better at reading the big picture.