Talk:Autonomous robot

From Wikipedia, the free encyclopedia

Autonomous robot is within the scope of WikiProject Robotics, an attempt to standardise coverage of Robotics. If you would like to participate, you can edit the article attached to this notice, or visit the project page, where you can join the project or contribute to the discussion.
B This article has been rated as B-Class on the Project's quality scale.
(If you rated the article please give a short summary at comments to explain the ratings and/or to identify the strengths and weaknesses.)
High This article has been rated as high-importance on the importance scale.
Peer review Autonomous robot has had a peer review by Wikipedia editors which is now archived. It may contain ideas you can use to improve this article.

Contents

[edit] Article redundant?

A robot is by definition autonomous. So this page is redundant. Any useful material should be moved to robot and this page deleted. Psb777 10:30, 24 Jan 2004 (UTC)

Even if you think I am wrong we cannot have a different page for every adjective-noun combination. bipedal robots, anyone? Psb777 10:34, 24 Jan 2004 (UTC)

I accept that I was wrong. The page should stay. Psb777 16:36, 8 Feb 2004 (UTC)

[edit] Please cite research

When people mention research, without citing it, or adding links, thery might as well not bother. Stan 14:59, 24 Jan 2004 (UTC)

[edit] Fully autonomous

I suggest this page be extended. Fully autonomous robots is a new field and definately merits some mention, in fact there's even already much controversy re: the darpa autonomous robot challenge 80.126.238.189 21:37, 3 Feb 2004 (UTC)

The name of the event is "DARPA Grand Challenge". It is not for autonomous robots but for autonomous ground vehicles. This is plainly obvious from the link added by User:80.126.238.189. And, in the context of this Talk page, seems a deliberately calculated misquote.

Further this field might be new but how new? Lots of noise but little info.

Psb777 23:21, 3 Feb 2004 (UTC)

Well, for starters, factory robots need not be very autonomous at all. They could simply follow a sequence of preprogrammed instructions. And they even usually need some amount of supervision.

Modern robots have been gaining more and more autonomy through time, but you may have noticed that even space probes still need some amount of control and guidence from the ground.

Making a robot *Fully* autonomous represents a kind of quantum leap. While the advances in robotics themselves are evolutionary, it does suddenly allow entirely new (revolutionary) applications.

I'm not sure when people first started working on fully autonomous robots, I think the wiki will sort that out in time. If not, I'm willing to research it.

Finally, hmm, well, you're being a bit pedantic. A robot is an automated mechanical device. An autonomous ground vehicle would definately fall under that definition.

80.126.238.189 00:44, 4 Feb 2004 (UTC)

Wikipedia often benefits from pedantry. My point was that I am saying robots are autonomous, autonomous robot is a tautology, and autonomous ground vehicle is not the same as autonomous robot but it is the same as robotic ground vehicle. But I am coming around to your POV: Why not have a page about fully autonomous robots? Psb777 10:01, 4 Feb 2004 (UTC)

Robots need not be autonomous

Hmm, I wish I could find a better link, but this will have to do: Autonomous Brains and Autonomous Robots. This article implies (IMHO correctly) that proper (fully) autonomous robots don't exist at this point in time, though robots that exhibit autonomy in some aspects of their behaviour do exist. The article isn't perfect, if nescesary perhaps I'll search some more later.

Google

Google [at the point in time when I looked] returns 205000 hits for "autonomous robot" and 35300 hits for "fully autonomous robot". Just from skimming some of the hits for autonomous robot, I found people take "autonomous robot" to mean among others:

  • A robot with one or more aspects that have been made autonomous. (A robot need not be autonomous at all by definition, it merely needs to be at least partially automated).
  • a fully autonomous robot (as an unreached goal?).
  • An autonomous(ly) mobile robot.

I'll admit that a more detailed search might be warrented, but it's a lot of material to wade through.

A washing machine is a non-autonomous robot

" Autonomous robot " need not be tautological at all. A modern washing machine is (per the definition on wikipedia) a robot. It is a mechanical device which has been automated. That means it contains a computer which can run multiple programmes, which are used to control the mechanism. It is not enough for a device to contain a computer. To consider something a robot, the mechanical parts of the device must be under computer control. For a washing machine this holds true.

While you can easily argue that a washing machine is automatic I don't think anyone would consider a washing machine to be autonomous.

You have to say why it's not autonomous. A washing machine cannot follow a map from A to B but the Mars Rover cannot wash and dry my clothes. Psb777 07:31, 8 Feb 2004 (UTC)
A washing machine requires assistence in the context of washing my clothes. It will perform certain steps automatically, but it cannot find my dirty clothes, wash them and hang them out to dry for me. I'm not sure if it's important yet, but note that a mars rover can more easily be modified to make it wash clothes than a washing machine can be modified to explore mars. Perhaps there's a certain quality that the mars rover has that the washing machine lacks? 80.126.238.189 11:57, 8 Feb 2004 (UTC)

My understanding of the term autonomous robot

AFAIK an Autonomous robot would be a robot that is capable of surviving on its own in a complex and possibly hostile environment. Such a robot would be autonomous in a similar sense to a biological organism being autonomous. This is a hard problem.

No that would be an explorer robot. It all hinges on the definition of autonomous. Psb777 07:31, 8 Feb 2004 (UTC)
I dare you to call an RQ-4_Global_Hawk simply an explorer robot. It looks like some (generous) amount of tweaking would allow people to put similar capabilites in commercial jetliners. 80.126.238.189 11:57, 8 Feb 2004 (UTC)

Fully autonomous robot?

"Fully autonomous robot" does sound a bit strange. We have been arguing that there are degrees of autonomy, but how true is this? Maybe something can't be just a little bit autonomous. Either it can be autonomus or it can't be autonomous. (If someone can argue convincingly that you *can* be just a little bit autonomous, I might change my mind on this.)

Autonomy is not a black and white quality. An ant is autonomous except under my boot. A washing machine is autonomous under my boot. Depends upon the desired task and the environment. Psb777 07:31, 8 Feb 2004 (UTC)

On the other hand Fully autonomous robot *can* be found on google.

Hmm, ok, I'll defer judgment on weather we need "fully autonomous robot". If people do decide on that, make "autonomous robot" a redirect to that page.

80.126.238.189 12:49, 4 Feb 2004 (UTC)

Either we must put all the robot text into this article or vice versa. Or someone should reply to my argument that this article is superfluous. Psb777 00:12, 4 Feb 2004 (UTC)


It's a subfield of robotics. Besides, that's what wiki-links are there for, to help prevent single monolithic pages. Finally, it looks like people have been editing the page since my last visit, a sure sign that the subject lives and will get more written about it over time. 80.126.238.189 00:44, 4 Feb 2004 (UTC)
Whoops, that was you. You definately improved the page. 80.126.238.189 00:53, 4 Feb 2004 (UTC)

Following a rewrite of big chunks of this and the addition of more material I have asked for Wikipedia:Peer_review. Psb777 00:31, 4 Feb 2004 (UTC)

In what way is a washing machine not autonomous? That a robot is necessarily autonomous is my (questionable?) POV, but an autonomous machine is not necessarily a robot. Or would others disagree about that too? Is any machine which operates without external guidance (and is therefore [somewhat?] autonomous) a robot? Or must some intelligence be displayed? But then we start having the debate being held over at artificial intelligence pages. As an aside there or on its talk page someone quotes someone as saying a thermostat has a belief system: It believes it is too hot, too cold or just right. Is a thermostat a robot? Is it autonomous? What is a robot? Psb777 14:15, 4 Feb 2004 (UTC)

I think maybe we need to define autonomous. It appears that autonomy is defined differently in different contexts. It depends on what is autonomous from whom or what.

So we'd have to look at that. What's autonomy in the real world?

After some googling, and also using some of my own knowlege, here's some aspects of what might be autonomy for some entity in the real world:

  • The ability to work for large amounts of time (months or years) on end without supervision.
  • The ability to repair itself without outside assistance.
  • The ability to adjust stratagies based on the surroundings.
  • The ability to adapt to surroundings without outside assistance.
  • The ability to travel from a randomly chosen point A on a map to a randomly chosen point B, without any outside assistance.
  • The ability to learn or gain new capabilities without outside assistance.
  • The ability to gain information about the environment by themselves.

Hmm, a key aspect of autonomy seems to be that such a system can basically look after itself and make descisions without outside assistance.

Such autonomous entities do exist. Think of living organisms, which cover all of the above. (To pre-empt one particular line of discussion: plants don't move themselves, but they can still reproduce and at least spread their genome from point A to point B. )

Robots don't cover most of the above. The darpa autonomous vehicle challenge covers just movement from the above list. Just achieving autonomous movement alone appears to be some kind of holy grail.

According to this definition I'm not sure how a robot could be " autonomous by definition ".

Hmm, perhaps you're using a different definition of autonomous. :-) Perhaps some of the hits on "autonomous robots" have some specific idea of what autonomy is in the context of building (real world?) robots. This might be useful to look up and/or figure out.

80.126.238.189 11:59, 5 Feb 2004 (UTC)

Autonomous means able to do some things automatically. That's all. And that is why all robots are autonomous. Psb777 07:31, 8 Feb 2004 (UTC)
You mean autonomous and automatic might be synonyms? I checked a thesaurus, and hmm, it does not appear to support such a statement. Apparently autonomous and automatic have different meanings.

Now that's interesting, robotic is sometimes used as a synonym for automatic.

Ok, is your position autonomous=automatic=robotic ? I do see! Well, according to dictionary.com, autonomous and automatic have different (even almost oppositional) meanings. So autonomous robot isn't a tautology, in fact it's more of a contradictio. Probably the person who coined the term knew this, and chose his/her wording to be deliberately inciteful (sic).

Will you at least acept the NPOV position that some people use the term autonomous robots as a designation for a certain class of robots? 80.126.238.189 11:57, 8 Feb 2004 (UTC)

Hmm, after some discussion here's a question: What happens if you move a factory robot to a new location, and assign it a new task?

Right. You need to reprogramme it. So a factory robot is only autonomous within the very strict confines of the location in the factory where it is working.

I guess that means the article needs to be boldly rewritten. 80.126.238.189 23:29, 7 Feb 2004 (UTC)

While doing detailed reading I found this bit: " even when used in space exploration is not much more than a radio controlled car if or when it is under direct human guidance, as is occasionally the case. "

this is simply wrong, the current nearest manned vehicles have a considerable radio lag between mars and earth. The vehicles *MUST* have some degree of autonomy, because there's a ~30 minute lag up and down. At no point in time can they be simply radio-controlled in real time. So the simply radio-controlled-cars bit is right out.

Note that Psb777 wrote here quickly with the intention of going through the vfd procedure. He did a great job with that. Now that the article is going to be kept, it's going to need some work though :-) 80.126.238.189 23:53, 7 Feb 2004 (UTC)

[edit] Capabilities of the Mars rovers

The Mars rovers MER-A and MER-B can navigate their own routes to destinations on the fly, by:

  • mapping the surface with 3-D vision
  • computing safe and unsafe areas on the surface within that field of vision
  • computing optimal paths across the safe area towards the desired destination
  • driving;
  • and repeating this cycle repeatedly until either the destination is reached, or there is no known path to the destination

They are also capable of running many tasks autonomously, such as sun-finding, automatic communications sessions, and auto-reboot. -- The Anome 23:59, 7 Feb 2004 (UTC)

But that is not an exhaustive list of the Mars Rover's capabilities. Some of the missed ones include "turn left" and "back up". Like a RC car. Psb777 07:31, 8 Feb 2004 (UTC)
Please explain. 80.126.238.189 11:57, 8 Feb 2004 (UTC)

What's your point? I agree, the Mars Rovers can indeed be programmed to perform pre-programmed operations just like an RC car -- however, this does not take away their autonomous navigation capabilities, which are what is relevant here. -- The Anome 12:29, 8 Feb 2004 (UTC)

[edit] Bold editing

I've done some bold editing on the article now. I'll leave it like this for a little while to see what happens next and what people do with it. 80.126.238.189 00:20, 8 Feb 2004 (UTC)

I've left in the part about factory robots being autonomous, but I'm not entirely sure how true that is. Aren't factory robots often set up in environments where variation is minimised? 80.126.238.189 00:26, 8 Feb 2004 (UTC)

[edit] Point proven: Article is redundant!

Everything can only cope with the environment for which it designed / adapted / evolved.

But what, then, is autonomous? Are you autonomous? If by autonomous you mean fully autonomous then nothing is. There is always some constraint. The Mars Rover is autonomous, say some, but it cannot manufacture its own ball bearings. The Sony robotic dog toys, say others, are autonomous but they cannot plug themselves into their battery charger. You are autonomous, I claim, but others deny the existence of your free will. And you cannot live on Mars. A factory automaton needs its factory to "survive".

Note my use of "automaton". We think of "automatons" as being less than "robots" yet, here, "autonomous robots" are uber-robots?

The terms robot and automaton are synonymous. And robotic and autonomous are synonymous. That was my orignal point.

Psb777 02:37, 8 Feb 2004 (UTC)

See above for my response on synonyms.

I think you are programmed to deny that all robots are autonomous, to some degree. -) Psb777 15:00, 8 Feb 2004 (UTC)

[edit] Humans fail the autonomy test?

Whereas I am capable of all the things listed as some of the things a fully autonomous robot might need to be able to do in the real world I note that none of my relatives are. Psb777 02:52, 8 Feb 2004 (UTC)

Well, let's look at that list and see which of the items your relatives might not be capable of.

  • The ability to work for large amounts of time (months or years) on end without supervision.

Well, with "work" here, I meant that they don't drop dead if no-one else looks after them. Their ability to do useful work without supervision might not show itself, but I'm sure they could be left unsupervised for long periods of time.

  • The ability to repair itself without outside assistance.

Human beings have an innate healing ability.

  • The ability to adjust strategies based on the surroundings.

Say for the problem of moving from point A to point B, if the chosen route is blocked, most humans will successfully adjust their route to avoid the obstacle.

  • The ability to adapt to surroundings without outside assistance.

I'm sure your relatives have moved house from time to time, or changed jobs. Or even if your relatives "moved here 60 years ago". The fact that they're still there means they've probably successfully adapted.

  • The ability to travel from a randomly chosen point A on a map to a randomly chosen point B, without any outside assistance.

Humans are capable of navigation. Untrained humans are less capable than trained humans, but on the long run even an untrained person is able to get quite far.

  • The ability to learn or gain new capabilities without outside assistance.

Human beings can learn new things from experience.

  • The ability to gain information about the environment by themselves.

Human beings have a sophisticated internal and external sensor system backed up with some of the best back end processing capabilities in the animal kingdom.


Okay, so with just this list I think your relatives have a good chance at being autonomous. Even so, biologists usually take into consideration that humans like to live in groups, so they might say that humans in general are usually autonomous on the society level .

80.126.238.189 12:38, 8 Feb 2004 (UTC)

I know you understand my comment was made in jest but the point made is valid: None of anybody's relatives is capable of all the things needed to qualify as fully autonomous (unless we considered them as a group). One may be incapable of finding work, another can't follow a map, another is confined to bed, another may be terminally ill. As someone said earlier, autonomy is a Holy Grail.

As I keep on trying to make others see, a robot which does not display any autonomy hardly seems to be a complete, full, fit robot at all. My washing machine has more intelligence and more autonomy than some of the programmed machine tools (numerically controlled machines) some here want to call industrial robots yet nobody is comfortable with calling my washing machine a robot. They would be happier calling it a robot if it had a 6 degrees of freedom arm (or preferably two arms) with which to stir the washing about even if it was less good at washing clothes!

Are we really going to define a robot as requiring an anthropomorphic feature, an arm, rather than autonomy. It is autonomy which is the defining attribute of a robot, not an arm.

These following two points seem true to me and are the ones I have been arguing:

  • Nothing is fully autonomous.
  • Anything which has no autonomy at all is not a robot.

The second point above is logically equivalent ( ~A => ~B is the same as B => A ) to this:

  • A robot has some autonomy.

It is not equivalent to this (but nevertheless this is a point with which I have some sympathy but I'm not really bothered enough about to argue):

  • Any (artificial) thing which is autonomous is a robot.

Anthropomorphism, the arm or the eye, is not an essential feature of a robot: Autonomy is.

Psb777 15:32, 8 Feb 2004 (UTC)

And that is why I think the page is redundant. As all robots are autonomous (to some degree) then autonomous robot is not required. If we start this article with This article is about highly autonomous robots or if the page is renamed then I would be happy.

Psb777 15:40, 8 Feb 2004 (UTC)

I think I understand your position. But what you seem to be doing is (accidentally) equating autonomous with automatic. I replied to an earlier comment by you in that sense. (Check your dictionary and /or thesaurus). dictionary.com is a useful resource.

80.126.238.189 16:10, 8 Feb 2004 (UTC)

I was confusing autonomous and automatic but I have stopped doing so. But my mistake was limited and did not, I believe, destroy my argument. More importantly it seems I am wrong about what people think robot means. I did strongly hold the belief that automatic was not enough to be robotic. I always believed autonomy was essential. Seems I am wrong. In future, when I would have used the unqualified word robot I will instead use intelligent robot or autonomous robot. I cannot change what appears to be the widely held view that a robot is any automated machine with an arm. The word robot means what people think it means. I give up! Or, rather, I accept my usage of the word robot was not common usage. Psb777 16:28, 8 Feb 2004 (UTC)

I think the term robot may also be used to describe Waldoes (non-autonomous remotely operated machinery. For instance, in the case of remote surgery, the surgeon may be said to be using a robot. —The preceding unsigned comment was added by WhiteDragon (talk • contribs) 20:41, 9 April 2004 (UTC)