Sevmorput was commissioned two years after the Chernobyl disaster. While she’s been laid up for long times, that wasn’t until the late 1990s and in the 2000s…
Oh, sigh. The paper documents how easy it is to spoof the pattern recognition algorithms used by these systems. That’s of interest to people with malicious intent. Very relevant to autonomous automobiles (e.g., a sticker that converts a stop sign into a banana) probably much less so for ships.
It is another example of the inherent limitations of what could be called “observational” sensor suites, those that look at naturally occurring inputs (ambient light) or returning “pings” from a transmitter (radar, lidar, sonar).
If you look at airborne collision avoidance systems, with one exception they all rely on active emitters, either a transmitter that squawks position periodically or transponders that squawk in response to being “painted” by an observational sensor. The only exception is ground proximity radar, which is tasked with detecting an object the size of the earth.
I’m quite sure the autonomous ship people will eventually end up in the same place, hopefully not through one or more casualties.
(As an aside, the need to have an active environment (transponders, fixed beacons, etc.) places an lower limit on the number of collisions that can be avoided by current autonomous automobiles. It will be interesting to see if that limit is above or below the societal tolerance for accidents.)
Cheers,
Earl
It will be interesting to see if warships will transmit their positions. I can see the United States not actively transmitting in contested waters (Persian Gulf, Yellow Sea, et al) which also happen to be crowded waters. This would put the burden on the warships to avoid (we know how well that works) or the autonomous ships to avoid non-transmitting ships who defy the rules of the road.
One way for the former to work better would be if autonomous ships transmitted their intentions / intended track with expected maneuvers along with their AIS data. Other ships would ‘know’ what the autonomous ships’ intentions were and could plan ahead.
IMHO having been a regular trader in the far east / PG / Singapore that is the only way that it can work. However, we are already in a two tier situation with class A AIS being mandatorily carried by vessels >300GT which does not include the hundreds of thousands under that all doing their own thing according to local custom and practice.
Eventually there may well be a ‘network of ships’ where communications / data / maneuvering are all connected but we, realistically, are probably thirty years away from that. Look at the 70 year old lakers!
Here’s how long it took in the aviation field:
http://www.eurocontrol.int/articles/history-future-airborne-collision-avoidance
Cheers,
Earl
With all that knowledge and experience from the Aviation industry it should be possible to develop a working system for ships that take away human error as much as possible.
Just like in the Aviation industry, where an aeroplane can take off, fly across oceans and land safely, using automated system, it should be possible for ships to do something similar. Initially with someone on board who can take over in case of failure, but eventually autonomously.
Drones, being operated remotely from the other side of the world, are able to spot and kill people with a reasonable accuracy. Why cannot a ship be remotely piloted into and out of ports from a nearby Port Control Tower?
If aircrafts can be protected from hackers taking control of them and turn them into weapons, or drones turned around and fire at own HQ, why is it IMPOSSIBLE to imagine that ships can be protected from similar attack?
Au contraire, it easy to imagine, but I have a good imagination, sometimes I think about the pretty waitress at that burger joint… but not going to happen. It’s the implementation that’s difficult.
More difficult to imagine is a flaw in almost every Intel chip made in the last 10 years
Today, three groups of security researchers from the Technical University of Graz, Cerberus Security, and Google Project Zero revealed a pair of defects in modern computers that allow adversaries to steal passwords and other sensitive data from virtually any computer in use today.
Didn’t see that coming.
Most aircraft are piloted with people in the cockpit. Even cargo planes fly with pilots. As others have repeadly said its a matter of human comfort rather than technology.
You seem to forget/ignore that we’ve had the ability to remote pilot aircraft for decades now. Hobbyists have been flying remote control airplanes for decades. Passenger aircraft have had autopilot for decades. We could have had pilotless cargo jumbo jets in the air decades ago. But we don’t.
And you forget that aircraft are highly unlikely to run into each other once aloft. That they fly in three dementions (bound by the ground below, the thin air above, and the rare mountain) and not sail in two dementions (bound by the common land) makes collisions many times less likely. Add to the fact that any person with a raft, paddle and a fishing pole can present a hazard to shipping but it’s still rare that any person can up and fly around - let alone fly around at 10,000m.
It’s hubris to expect the matter of autonomous ships will be solved and IMPLIMENTED beyond demonstrator technology any time soon.
God I’m tired of this shit. Look, here’s my modest proposal:
As backyard hobbiests have been flying remote controlled kerosene powered airplanes from their backyards for scores of years now. 1950s maybe?
As jumbo jets have had autopilot for scores of years and radio direction finding would be fine over land to adjust courses.
As jumbo jets arrive/depart from large airports.
Therefore train hobbiests to takeoff and land jumbo jets from large airports and let autopilot fly between airports via radio beacon ‘roads’ to follow.
Oh, we did have unmanned aircraft back then too. For example, the DH.82 Queen Bee (1935) or the truly ‘autonomous’ aircraft AIM-7 Sparrow (1959).
This is all 1950s tech here. SO WHY DO WE STILL HAVE PILOTS ON AIRCRAFT?
I didn’t see that flaw, not only Intel chips but just about every other chip too, coming.
That will affect both the aviation and military use of those chips as much, if not more, than any autonomous ship of the future, will it not??
By the time autonomous ships becomes common, I hope that problem have been solved and plugged.
It may be that according to surveys of airline passengers, it’s because they want warm bodies in the cockpit but I’m not convinced the current technology would have brought US Airways 1549 to a successful ditching saving the lives of all passengers.
So the guys who play with little remote controlled boats on a pond could bring ships into harbor and dock them?
Dude that’s gotta be some prime bud you’re smoking.
The point I was making with my ‘modest proposal’ was that the technology for pilotless aircraft existed sixty years ago. And yet, sixty years later, we still don’t see autonomous aircraft except for military applications and the occasional demonstrator projects.
Our Norwegian friend seems to think that just because the technology for crew-less ships exists it’s implementation is imminent. Nuclear cargo ships and supersonic passenger transport these are extant technologies. So why aren’t the seas and skies full of them?
Just because the technology exists for crew-less ships doesn’t mean we will suddenly have crew-less ships everywhere.
Your Norwegian friend is a realist and do not think that commercial crewless ships will be crossing oceans in the next decade or more, which I have stated repeatedly.
It is not because the technology to do so isn’t available, but because it will take at least that long to change the laws, rules and regulations to allow that to happen.
Yes I do think that short sea shipping within restricted areas will happen within that time though. I also think it will happen first in Scandinavia, followed by Europe, China and Japan.
As for US mariners; you should feel safe that there will still be old and obsolete ships to serve on. There are no signs that will change in the foreseeable future.
Gotcha
Trains.
How complicated can driving a train be? It runs on rails using technology over a hundred years old! If ever there was a lower hanging fruit for autonomous operation then an engineer-free train I don’t know what it would be. And yet from Japan to Europe to the United States the fright trains and passenger trains still operate with on-board human engineers.
Sure, you can find a light rail here and there without drivers. But even some of Singapore’s MRT lines still have drivers. (And it’s not as if their MRT is a challenging environment for automation. Many basement model train aficionados have more complex system environments then Sing MRT.)
All this autonomous ship hysteria must be familiar to days past when every town would have fossil-free nuclear power plants, cars would fly, and travel between continents would be on plush aircraft at twice the speed of sound.
Which do you think will happen last?
-
Commercial flights from New York to Los Angeles will take less than two hours.
-
Polio will be eradicated from the face of the Earth.
-
Autonomous ships over 16,000 GRT will outnumber manned ships over 16,000 GRT.
-
You and I will both be dead.
Not only MRT trains are operated driverless.
Large mining operations are getting close to autonomous operation:
This happens even in Australia, where the Unions are strong and sometimes militant.
We still have pilots because of human comfort.
If I’m in one of these examples of human inventiveness I take comfort in the thought that the guy in the pointy end is just as keen to get his or her arse back on the deck as I am.