If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.

(Gizmodo)   Someday this robot will run faster then all of us, catch Sarah Connor   (gizmodo.com) divider line 36
    More: Scary, IEEE, University of Tokyo, blades, actuators, pet projects  
•       •       •

4198 clicks; posted to Geek » on 13 Dec 2010 at 10:25 PM   |  Favorite    |   share:  Share on Twitter share via Email Share on Facebook   more»



36 Comments   (+0 »)
   

Archived thread
 
2010-12-13 08:37:28 PM
www.jdmfilmreviews.com

Approves
 
2010-12-13 09:56:42 PM
Currently, Niiyama and his team are busy trying to teach the robot how to run, which involves analyzing the timing and patterns in which human muscle sets contract and relax. Right now, Athlete can only cover about three steps (at 1.2 meters/sec) before collapsing

It cant be too hard if Dumb Luck built it the first time.
 
2010-12-13 10:14:47 PM
Bevets: Currently, Niiyama and his team are busy trying to teach the robot how to run, which involves analyzing the timing and patterns in which human muscle sets contract and relax. Right now, Athlete can only cover about three steps (at 1.2 meters/sec) before collapsing

It cant be too hard if Dumb Luck built it the first time.


Bevets, I'd like to introduce you to "the world outside of your house."

Natural Motion- Biped Learning to Walk Using Evolved Neural Nets
 
2010-12-13 10:26:51 PM
Can they build a robot that will teach people the difference between then and than?
 
2010-12-13 10:27:12 PM
bigdavediode:

Natural Motion- Biped Learning to Walk Using Evolved Neural Nets

I wonder why Dr Niiyama and team didnt just borrow this simulation? It would have saved them a lot of work.
 
2010-12-13 10:30:29 PM
Bevets: I wonder why Dr Niiyama and team didnt just borrow this simulation? It would have saved them a lot of work.

Maybe because the robot they're creating IS MISSING A LOWER LIMB.

God never meant for you to be this dumb. You have a brain, start using it.
 
2010-12-13 10:32:39 PM
Then?
 
2010-12-13 10:33:02 PM
(Well, technically it's missing both lower limbs.)
 
2010-12-13 10:36:43 PM
That Big Dog thing linked was an interesting watch.

http://gizmodo.com/368651/new-video-of-bigdog-quadruped-robot-is-so-stunning-it s -spooky
 
2010-12-13 10:40:50 PM
whatshisname: Can they build a robot that will teach people the difference between then and than?

I think subby means that the robot will first run faster, then all of us will catch Sarah.
 
2010-12-13 11:14:01 PM
bigdavediode: (Well, technically it's missing both lower limbs.)

Do yourself, and the rest of us, a favour and stop replying to it. Kind of a "the only way to win is to not play" sort of thing.
 
2010-12-13 11:25:54 PM
On the bright side once our robot sprinting overlords come I don't have to be faster than they are. I only have to be faster than Sarah Connor.

Also speaking of robots I always thought Bevets was an automated bot. Maybe it has... evolved. It can actually quote people now. Though I have yet to see it answer a direct question...

Science can only do so much. At least for now.
 
2010-12-13 11:27:18 PM
Then when?
 
2010-12-13 11:32:15 PM
If it's controlled with q, w, o, and p then I don't think we have too much to worry about
 
2010-12-13 11:33:04 PM
It's never going to learn unless you take away the safety wire. As long as it thinks that it is safe from injury, it will never learn to run for itself.
 
2010-12-13 11:36:12 PM
Mountain Doo: If it's controlled with q, w, o, and p then I don't think we have too much to worry about

I see a new show, it's like ow my balls, but less sacking more laughing at tard like inability to run.
 
2010-12-14 12:28:54 AM
Came here to make sure someone told subby of the most odious of grammatical errors.

/ leaving slashy
 
2010-12-14 02:21:45 AM
i was hoping it would get further than nowhere, but soon. soon.

this stuff isn't even needed... once the AI is better, robots will be deadly enough on wheels and in the air. If they need legs, they will design them themselves.
 
2010-12-14 02:27:05 AM
 
2010-12-14 04:29:11 AM
Am I the only one who sort of feels bad for the robot when it falls down?
 
2010-12-14 06:10:21 AM
I thought this was going to be a cool robot thread. Now I see it's just going to be another stupid Bevets thread :-(
 
2010-12-14 07:32:41 AM
picturescrazy: Am I the only one who sort of feels bad for the robot when it falls down?

No, no, it's a necessary way to build up its resentment towards humanity.

You can help the process by loudly exclaiming, "HA HA!" into your computer's microphone.
 
2010-12-14 08:40:52 AM
Bevets: It cant be too hard if Dumb Luck built it the first time.

I would say there are three things that limit the application of the principals of evolution to the improvement of this robot:

1. The rate of change would have to be very slow, otherwise you might skip over the optimal configuration. In practical terms, this means that you'd have to build a lot of robots - like millions of them, because you need several robots for each generation in order to have variation, and you need a great many generations in order to actually evolve. It's simply impractical to build that many robots.

2. You could use computer simulation to avoid building the robots, but I imagine they actually did that. I'm sure the code the wrote works just fine in the simulation, but any tiny difference between simulation and reality, for example a minute static charge in one of the motors, causes the robot to fall. Furthermore, what most people don't realize is that a genetic algorithm isn't exactly the same as evolution, any more than LIFO is exactly the same as stacking plates. A genetic algorithm finds an optimal set of parameters, but it doesn't actually modify its own code. I suppose there's nothing stopping you from writing one that way, but most people just need to find an optimal set of parameters, so they don't write them that way. People hear the term though, and they assume it's the same as evolution in nature.

3. The final, and perhaps most important reason, is that while we often look at how living things have evolved and we marvel at what appears to be great ingenuity, in fact the processes of evolution too frequently yield sub-optimal designs. There is a lot of vestigial crap in a naturally-evolved organism. We can actually do better than that when we design systems ourselves. For example, consider the human eye. Our photo receptors are installed backwards. The nerves leave the cell at the front of the eye, then travel along it periphery to get to the back where they are bundled up with other nerves. This is not the optimal design. It means that a lot of room in the eye is wasted on nerves instead of cells. The eye of the octopus is built the "right" way, with the nerves coming from the back of the cells. As a result, an octopus eye can have a higher resolution than a human eye.

Why is our eye built this way? Why hasn't evolution corrected this deficiency? It's called an evolutionary cul de sac. Without getting too technical, it's what happens when you have a local optimum that isn't "the" optimum. They have this problem with genetic algorithms too, and in all honesty it can happen with intelligently designed systems as well - it's why we use qwerty keyboards and english measuring units. Anyway, the point is, if you tough it out and design a robot, you should end up with a better robot that makes more efficient use of materials than if you try to evolve one.

I'm sorry that people respond to you reflexively. I think this is an interesting subject and that it's only fun to post when you disagree with someone.
 
2010-12-14 09:53:49 AM
bigdavediode:

Natural Motion- Biped Learning to Walk Using Evolved Neural Nets

Bevets:

I wonder why Dr Niiyama and team didnt just borrow this simulation? It would have saved them a lot of work.

Tofu:

You could use computer simulation to avoid building the robots, but I imagine they actually did that. I'm sure the code the wrote works just fine in the simulation, but any tiny difference between simulation and reality, for example a minute static charge in one of the motors, causes the robot to fall. Furthermore, what most people don't realize is that a genetic algorithm isn't exactly the same as evolution, any more than LIFO is exactly the same as stacking plates. A genetic algorithm finds an optimal set of parameters, but it doesn't actually modify its own code. I suppose there's nothing stopping you from writing one that way, but most people just need to find an optimal set of parameters, so they don't write them that way. People hear the term though, and they assume it's the same as evolution in nature.

Tofu:

The final, and perhaps most important reason, is that while we often look at how living things have evolved and we marvel at what appears to be great ingenuity, in fact the processes of evolution too frequently yield sub-optimal designs. There is a lot of vestigial crap in a naturally-evolved organism. We can actually do better than that when we design systems ourselves. For example, consider the human eye. Our photo receptors are installed backwards. The nerves leave the cell at the front of the eye, then travel along it periphery to get to the back where they are bundled up with other nerves. This is not the optimal design. It means that a lot of room in the eye is wasted on nerves instead of cells. The eye of the octopus is built the "right" way, with the nerves coming from the back of the cells. As a result, an octopus eye can have a higher resolution than a human eye.

We are far from understanding the complexity of individual organisms, let alone the entire ecosystem in which that organism lives. What appears to be less than optimal design to us with our limited knowledge may actually be an optimal design when the entire system is considered. Consider the thickness of armor plating on the side of a warship. Since the purpose of such plating is to protect the ship from the puncture of an incoming warhead, it is advantageous to make the plating as thick as possible. Yet the plating on actual warships is much thinner than it could be made. The reason is, of course, that an increase in plating thickness makes the ship heavier, and thus slower. A less movable ship is more likely to get hit more often and less likely to get to where it is needed when it is needed. The actual thickness of the armor on a warship is a tradeoff -- not so thin as to make the ship too easily sinkable, and not so thick as to make the ship too slow. We know too little about the complexity of organisms and the environment in which they live to conclude that any one particular feature is actually less than optimal. ~ Kurt Wise
 
2010-12-14 10:10:01 AM
Bevets: What appears to be less than optimal design to us with our limited knowledge may actually be an optimal design when the entire system is considered.

Quite true, and this is precisely why I gave you two examples, one was the opposite of the other. My point was that in terms of what is optimal, they are mutually exclusive. They can't both be the best possible design, because they're exactly the opposite. If you say the human eye is perfectly designed, then I'm left wondering why the octopus eye is better.

Unless you can propose even a hypothetical reason why they might both be optimal, I will continue to believe that the human eye is very slightly flawed in this one particular design element. I realize one is used underwater and the other is not - but this isn't a reason to handicap the human eye. I maintain that the human eye would be improved by positioning the photo receptors the other way around - like the eye of an octopus. I hope you can understand that as an atheist, I will not be satisfied by an explanation that says, *shrug* "I guess god had a reason to do this." To me, it is clearly flawed.

And there are many other examples too, in humans and otherwise, of leftover dregs from the evolution of our species.
 
2010-12-14 12:12:30 PM
Tofu:

The final, and perhaps most important reason, is that while we often look at how living things have evolved and we marvel at what appears to be great ingenuity, in fact the processes of evolution too frequently yield sub-optimal designs. There is a lot of vestigial crap in a naturally-evolved organism. We can actually do better than that when we design systems ourselves. For example, consider the human eye. Our photo receptors are installed backwards. The nerves leave the cell at the front of the eye, then travel along it periphery to get to the back where they are bundled up with other nerves. This is not the optimal design. It means that a lot of room in the eye is wasted on nerves instead of cells. The eye of the octopus is built the "right" way, with the nerves coming from the back of the cells. As a result, an octopus eye can have a higher resolution than a human eye.


Bevets:

We are far from understanding the complexity of individual organisms, let alone the entire ecosystem in which that organism lives. What appears to be less than optimal design to us with our limited knowledge may actually be an optimal design when the entire system is considered. Consider the thickness of armor plating on the side of a warship. Since the purpose of such plating is to protect the ship from the puncture of an incoming warhead, it is advantageous to make the plating as thick as possible. Yet the plating on actual warships is much thinner than it could be made. The reason is, of course, that an increase in plating thickness makes the ship heavier, and thus slower. A less movable ship is more likely to get hit more often and less likely to get to where it is needed when it is needed. The actual thickness of the armor on a warship is a tradeoff -- not so thin as to make the ship too easily sinkable, and not so thick as to make the ship too slow. We know too little about the complexity of organisms and the environment in which they live to conclude that any one particular feature is actually less than optimal. ~ Kurt Wise

Tofu:

Quite true, and this is precisely why I gave you two examples, one was the opposite of the other. My point was that in terms of what is optimal, they are mutually exclusive. They can't both be the best possible design, because they're exactly the opposite. If you say the human eye is perfectly designed, then I'm left wondering why the octopus eye is better.

Unless you can propose even a hypothetical reason why they might both be optimal, I will continue to believe that the human eye is very slightly flawed in this one particular design element. I realize one is used underwater and the other is not - but this isn't a reason to handicap the human eye. I maintain that the human eye would be improved by positioning the photo receptors the other way around - like the eye of an octopus. I hope you can understand that as an atheist, I will not be satisfied by an explanation that says, *shrug* "I guess god had a reason to do this." To me, it is clearly flawed.

And there are many other examples too, in humans and otherwise, of leftover dregs from the evolution of our species.




Octopod eyes are not set up for the perception of small detail, but for the perception of patterns and motion thus eliminating the need for the very high processing power seen in human and other vertebrate eyes. ~ Sean Pitman

Would hundreds of thousands of vertebrate species -- in a great variety of terrestrial, marine, and aerial environments -- really see better with a visual system used by a handful of exclusively marine vertebrates? In the absence of any rigorous comparative evidence, all claims that the cephalopod retina is functionally superior to the vertebrate retina remain entirely conjectural. ~ George Ayoub

I am glad you are a man of details, because I have been looking for a detailed description of the origin of life. Normally I get a vague outline but almost no details. Atheists tell me they are 'working on it'. Do you have a detailed recipe? You can warm up with details on the bacterial flagellum if you like.
 
2010-12-14 12:36:37 PM
Tofu, you've been beaten by Bevets. You approached the issue of octopus eyes versus human eyes because of "superiority" which is subjective. All he has to do is to claim that the octopus eye is suboptimal, which he has. So what if humans have a blind spot in the middle of their eyes and octopi don't -- all he has to do is to deny that this is a disadvantage. There is nothing that he can't deny.

Once you're on subjective, unscientific ground you're going to be dominated by Bevets. Irrationality is his territory.
 
2010-12-14 12:41:41 PM
Bevets: I am glad you are a man of details, because I have been looking for a detailed description of the origin of life. Normally I get a vague outline but almost no details. Atheists tell me they are 'working on it'. Do you have a detailed recipe? You can warm up with details on the bacterial flagellum if you like.

Scientists have created amino acids and protobionts spontaneously by just letting chemicals interact.
 
2010-12-14 01:49:31 PM
Is the robot controlled by Q and W for its thighs, and O and P for its calves?
 
2010-12-14 01:51:20 PM
Mountain Doo: If it's controlled with q, w, o, and p then I don't think we have too much to worry about

And of course I failed to spot this comment as I skimmed.
 
2010-12-14 04:35:54 PM
Bevets: Octopod eyes are not set up for the perception of small detail, but for the perception of patterns and motion thus eliminating the need for the very high processing power seen in human and other vertebrate eyes.

I don't doubt that this is true, but it doesn't speak to the single characteristic under discussion. It's more likely a description of their focal length, for example. What I brought up is a specific difference between human an octopus eyes. In this specific characteristic, the design of our eye is flawed. The "correct" design is what the octopus has.

Bevets: Would hundreds of thousands of vertebrate species -- in a great variety of terrestrial, marine, and aerial environments -- really see better with a visual system used by a handful of exclusively marine vertebrates?

This too misses the point. This is not a question of their "visual system" or ours. Our visual system is demonstrably adequate.

Rather, this is a question of one small part of the design.

Look, it's as if I'm trying to find a reason to criticize an iPhone. So I say, "it only has a 5MP camera - higher resolution cameras are possible in a phone - the HTC Evo has an 9MP camera." Then you say, "the iPhone is made for hip Apple people." (this quote, I think, is parallel to your Pitman quote above. It references the whole phone in the same way that Pitman referenced the whole eye). I don't doubt what the iPhone is for. I'm simply saying that in this one small characteristic, the Evo is better. Then you say, "would the iPhone necessarily be a better phone if it had the camera from an Android? Who knows!" (this, I think, is parallel to your Ayoub quote in that it attempts a qualitative assessment of the entire phone).

Bevets: I have been looking for a detailed description of the origin of life.

To my knowledge there is no detailed description. Furthermore, this isn't a question about evolution. Evolution is concerned with what happened after life began, not how it began. How it began is, quite honestly, a very intriguing mystery. There are clues. We can see the stuff of life forming in vast interstellar clouds, and even create it ourselves in experiments designed to simulate the early Earth. But I don't think that anyone can honestly claim to know for sure how it happened. We haven't yet even managed to "boot up" a simple cell from non-living parts, though Greg Ventor claims to be close. It's a real frontier of science.

My favorite origin theory involves a contribution of crystals in clay in ancient stream beds. I forget who came up with this theory.
 
2010-12-14 06:13:18 PM
Tofu: To my knowledge there is no detailed description. Furthermore, this isn't a question about evolution. Evolution is concerned with what happened after life began, not how it began. How it began is, quite honestly, a very intriguing mystery. There are clues. We can see the stuff of life forming in vast interstellar clouds, and even create it ourselves in experiments designed to simulate the early Earth. But I don't think that anyone can honestly claim to know for sure how it happened. We haven't yet even managed to "boot up" a simple cell from non-living parts, though Greg Ventor claims to be close. It's a real frontier of science.

Now he'll commit argument from ignorance -- "See? Science doesn't know!"
 
2010-12-14 08:58:57 PM
Tofu:

Our photo receptors are installed backwards. The nerves leave the cell at the front of the eye, then travel along it periphery to get to the back where they are bundled up with other nerves. This is not the optimal design. It means that a lot of room in the eye is wasted on nerves instead of cells. The eye of the octopus is built the "right" way, with the nerves coming from the back of the cells. As a result, an octopus eye can have a higher resolution than a human eye.

Bevets:

We are far from understanding the complexity of individual organisms, let alone the entire ecosystem in which that organism lives. What appears to be less than optimal design to us with our limited knowledge may actually be an optimal design when the entire system is considered. Consider the thickness of armor plating on the side of a warship. Since the purpose of such plating is to protect the ship from the puncture of an incoming warhead, it is advantageous to make the plating as thick as possible. Yet the plating on actual warships is much thinner than it could be made. The reason is, of course, that an increase in plating thickness makes the ship heavier, and thus slower. A less movable ship is more likely to get hit more often and less likely to get to where it is needed when it is needed. The actual thickness of the armor on a warship is a tradeoff -- not so thin as to make the ship too easily sinkable, and not so thick as to make the ship too slow. We know too little about the complexity of organisms and the environment in which they live to conclude that any one particular feature is actually less than optimal. ~ Kurt Wise

Tofu:

I hope you can understand that as an atheist, I will not be satisfied by an explanation that says, *shrug*

Bevets:

Octopod eyes are not set up for the perception of small detail, but for the perception of patterns and motion thus eliminating the need for the very high processing power seen in human and other vertebrate eyes. ~ Sean Pitman

Tofu:

This is not a question of their "visual system" or ours. Our visual system is demonstrably adequate.

Rather, this is a question of one small part of the design.

Look, it's as if I'm trying to find a reason to criticize an iPhone. So I say, "it only has a 5MP camera - higher resolution cameras are possible in a phone - the HTC Evo has an 9MP camera." Then you say, "the iPhone is made for hip Apple people." (this quote, I think, is parallel to your Pitman quote above. It references the whole phone in the same way that Pitman referenced the whole eye). I don't doubt what the iPhone is for. I'm simply saying that in this one small characteristic, the Evo is better.


Bevets:

Would hundreds of thousands of vertebrate species -- in a great variety of terrestrial, marine, and aerial environments -- really see better with a visual system used by a handful of exclusively marine vertebrates? In the absence of any rigorous comparative evidence, all claims that the cephalopod retina is functionally superior to the vertebrate retina remain entirely conjectural. ~ George Ayoub

Tofu:

Then you say, "would the iPhone necessarily be a better phone if it had the camera from an Android? Who knows!" (this, I think, is parallel to your Ayoub quote in that it attempts a qualitative assessment of the entire phone).

Do you know?

I do not agree that the comparison of different phone brands is a fitting analogy to the comparison of a human with an octopus. Are you suggesting that God is limited to creating ONLY one optimal organism? If not please explain.

Bevets:

I am glad you are a man of details, because I have been looking for a detailed description of the origin of life. Normally I get a vague outline but almost no details. Atheists tell me they are 'working on it'. Do you have a detailed recipe? You can warm up with details on the bacterial flagellum if you like.

bigdavediode:

Scientists have created amino acids and protobionts spontaneously by just letting chemicals interact.

Tofu:

To my knowledge there is no detailed description. Furthermore, this isn't a question about evolution. Evolution is concerned with what happened after life began, not how it began. How it began is, quite honestly, a very intriguing mystery. There are clues. We can see the stuff of life forming in vast interstellar clouds, and even create it ourselves in experiments designed to simulate the early Earth. But I don't think that anyone can honestly claim to know for sure how it happened. We haven't yet even managed to "boot up" a simple cell from non-living parts, though Greg Ventor claims to be close. It's a real frontier of science.

My favorite origin theory involves a contribution of crystals in clay in ancient stream beds. I forget who came up with this theory.


Every time I write a paper on the origin of life, I determine I will never write another one, because there is too much speculation running after too few facts. ~ Francis Crick

Next time somebody tells you something that sounds important, think to yourself, 'Is this the kind of thing that people probably know because of evidence or is it the kind of thing that people only believe because of tradition, authority or revelation?' And next time somebody tells you that something is true, why not say to them, 'What kind of evidence is there for that?' And if they can't give you a good answer, I hope you'll think very carefully before you believe a word they say. ~ Richard Dawkins
 
2010-12-14 09:42:40 PM
Bevets: Every time I write a paper on the origin of life, I determine I will never write another one, because there is too much speculation running after too few facts. ~ Francis Crick

Argument from ignorance.

There you go, Tofu. I'm not sure why Bevets felt the need to be so wordy about it, though.
 
2010-12-14 11:53:45 PM
Bevets: Are you suggesting that God is limited to creating ONLY one optimal organism? If not please explain.

No. I'm saying

a: that evolution has produced many designs for eyes.

b: that the human eye is good enough, that it affords us selective advantage, but that its design is not at all optimal. That there is no good reason for the photo receptors to have their nerve fibers exit from the front of the eye. That if you started from a blank sheet of paper designing an eye for a human, you would not design it this way.

c: that one of the failings (or perhaps limitations is a better word) of evolution is that it is susceptible to this kind of cul de sac, where fixing this design flaw requires such a large change in the organism, for so little immediate advantage, that it is unlikely evolution will ever "fix" it.

A hypothetical god can do whatever he wants, but if you believe there is a god, and you believe that the human eye is perfectly designed, then how do you explain the issue with the photo receptors? Why is this way a better design than the octopus way?

Also Bevets, I have to tell you, I'm not actually clicking your links. I would click them if there were fewer of them. If each post contained maybe one link, I might think, "oh this link must be important" but your posts contain dozens of links. Hell, you're quoting me and *adding* links into my posts! So instead of thinking, "this must be important" I'm just overwhelmed and I end up clicking nothing at all.

And finally, I'm happy to respond to you, and if you quote someone else I may respond to that too, but it's much less likely. It's just not as interesting. In your last post, you quote Dawkins. Well, I don't disagree with any particular part of that quote. So, I don't know what to do with it. It doesn't seem to be making any point at all. I just shrug and move on. So I guess what I'm saying is that you're sort of wasting your time with the excessive links and quotes. I enjoy talking to you, but I'm skipping over the other stuff you're posting.
 
2010-12-15 03:48:44 PM
oh ffs people. I thought I told you to stop feeding the damn thing.
 
Displayed 36 of 36 comments



This thread is closed to new comments.

Continue Farking
Submit a Link »





Report