PDA

View Full Version : [Interesting] new humanoid robot, what do you think of it



openmindedjjj
09-04-2008, 02:53 AM
ok so i just saw this video on youtube and i thought you guys might like it. its a new robot and i would totally buy it.. i noticed that the video is almost 2 years old so i tried to find some new news on the robot but i was unable to find any new or current info on him. if you guys can find some new information on weather he would be going for sale or any other stouf feel free to write.
YouTube - Mahru Humanoid Robot

Adrenalynn
09-04-2008, 03:21 AM
Is this like you wanting to buy an Asimo for a couple grand?

You did hear the ~$200,000 in the video, right? You have two hundred thousand dollars laying around to buy a slow basically useless 'bot? ;)

4mem8
09-04-2008, 03:46 AM
One has to wonder Eh!! Wish I had two hundred grand to spare. I could build a zillion Mechs

sthmck
09-04-2008, 09:04 AM
I am pretty sure I wouldnt spend my 2 hundred grand on that. I like how it said they would like to have them in homes doing simple chores by 2008.

Layton
09-04-2008, 03:08 PM
Samsung and KIST have already developed newer versions in this series, you can read about them and watch videos (at the bottom of each entry) on my blog:
Mahru & Ahra (http://robotbling.gametrailers.com/gamepad/index.php?action=viewblog&id=326657)
Mahru II & Ahra II (http://robotbling.gametrailers.com/gamepad/index.php?action=viewblog&id=327161) & Mahru III (http://robotbling.gametrailers.com/gamepad/index.php?action=viewblog&id=327161)
The last I heard was Mahru III was still in development as of early 2007, but nothing since then. It's possible the project is still going though.

Matt
09-04-2008, 03:14 PM
Thanks for the video openmind!


"It can deliver drinks in 10 years"

Allow me to rant :D These kind of robot projects I think are massive wastes of time, money, and energy. It's trying to run before you can crawl or even roll over. It's gross misapplication of technology and a horrifyingly bad business model. Sinking tens of millions into such far fetched ideas such as personal humanoid helper robots for the home is pure lunacy. Let's say this hits the market in 8 years for $200,000. A minimum wage worker may cost $12/hr. 200000/12 = 16,666 hours worth of work (8 years!) Hmmm, which is a better value, a full time worker for 8 years or a robot that can hardly pick up a glass and put it away and weighs so much that it would kill your child if it fell on him/her??? LOL It's like trying to design space shuttles for "the average family."

I'm not missing the point about research. Of course we always sink millions into that far ahead of a technologies readiness for market and that's cool. But just don't go spouting on about how you're going to turn it into a product.

I still think the money is better spent on tangible things like vision recognition systems, navigation, improved actuators & locomotion. How can we build a humanoid when we haven't even perfected the foot or eyeball? Waaaaaay ahead of themselves over there, may as well have a bonfire with your investors money.

/rant lol

Anyone disagree? Feel free argue the other side :)

Layton
09-04-2008, 03:25 PM
I disagree with you. There's no indication that they have any real plans of ever commercializing this robot. The robot is being developed by KIST which is a research institute not a company. Secondly, you are not giving the developers their due credit. You say money would be better spent on image recognition and stuff, which is precisely what these developers are doing. :/

I suppose you feel the same way about Toyota's Partner Robot program? they are already testing a partner robot in hospitals this year. The more expensive robot projects being developed in Japan are likely much more complex than anything you have experience with; perhaps you're out of your depth, or simply envious of what they are doing?

Oh, I just noticed who you are Matt! :) Sorry if I come across kind of abrasive, but I find it even more amazing that you would thumb your nose at projects like the Japanese and Korean humanoids when those very projects (be it QRIO, Asimo, Toyota's robots, Mahru, Hubo, etc) are the very ones that are driving interest in robotics in the mainstream, and helping to grow your own business! :roll eyes:

lnxfergy
09-04-2008, 04:08 PM
Personally I'm more disgusted with all these "humanoid" robot projects. These projects generally don't generate lots of great vision/locomotion/etc research. It takes so much time to develop the legs and stability most projects run out of money before they get to the vision/speech/thinking/etc research. Why do we need legs? A wheeled robot with a human-shaped torso is every bit as humanoid as a walking one - or are you gonna argue that people in wheelchairs not as human as those who can walk? Nearly everyone of these robots is based on some form of ZMP, which is inherently stable, but slow and useless in the real world. These walking bots are even worse than wheeled bots on such real world terrain as stairs and grass... Walking bots also use more power, most of them are on a tether, or only run for 30 minutes. We've seen dozens of asimo-type robots over the years, and the only ones that do anything interesting are typically R/C controlled. Lets get some real intelligence and sensing first, then work on making it look cool.

And as to business side, we've been saying robots to clean the house and do your chores are "just around the corner" for almost 50 years - give it a rest. These projects get the public hyped up - and then let them down... Even robots like the AIBO, whose dog form-factor inherentily lowers the expectations of the end user, is a dissappointment. Too much research is going into funky locomotion like walking, and too little into the important stuff - sensing/thinking.

-Fergs

darkback2
09-04-2008, 04:49 PM
I believe they said they had invested 200k...not that it would cost 200k...

I agree with matt to some degree...The point is, that a 1.5 meter tall humanoid costs a ton to make, but doesn't offer much in the way of advantages over a .5 meter tall humanoid...or .25 for that matter...That said, robots like Asimo, and all the other hugamoids are great show peices...they attract peoples interest. Especially when you cut out all of the falling down. That said, the money would probably be better spent developing all of the other parts such as 3D vision systems, navigation sensors and whatnot.


http://www.youtube.com/watch?v=B7QEU4GDrUk

sam
09-04-2008, 05:11 PM
Oh, I just noticed who you are Matt! :) Sorry if I come across kind of abrasive, but I find it even more amazing that you would thumb your nose at projects like the Japanese and Korean humanoids when those very projects (be it QRIO, Asimo, Toyota's robots, Mahru, Hubo, etc) are the very ones that are driving interest in robotics in the mainstream, and helping to grow your own business! :roll eyes:

Hello.

Honnestly I don't think that these kinds of project specialy help for the interest in the general public. I find it very interesting/awesome that they can make a robot walk like this. But I talked to all my friends and family relatives. None of them knew ASIMO or other robots that were mentionned before in this thread.

Each time I show them the robots, they ask soemthing like : "Can it clean the house" or "What can it do". Well, you can't lie that it will be able to clean the house, and even if it could magicly be able to do it tomorow, the general public could/would (I think they would buy a house before) buy one. Not of much utility. As for waht it can do, the list includes walking, running, avouding people sometimes... A lot of people to wich I talk about this are generaly dispointed in the time, effort and money given to these thigs when all that comes out is that it can walk. Well of course they don't know how complicated it possibly could be (I found out that it was hard doing robots when I tried to tie my first wheel to a motor...) to build a robot.

I beileve too that a robot with a wheeled base would be better on focusing the money and effort into what I think is the important stuff. Walking for later.

Adrenalynn
09-04-2008, 05:15 PM
Wow! You Go Fergs! Even better than Matt's post.

I've seen my hobby LIDAR meet with derision until I slapped it on a base that moved around. It was "nothing" until the brains were slapped on a base that a 5 yo could throw together. Oooh - now it's important.

We should be working on the really DIFFICULT problems that will yield real results. The problems that don't have discoverable optimal solutions but are np-complete and require tremendous brainpower to arrive at iteratively better approaches. Navigation is the biggie, of course - look at just the last few years of DARPA sponsored research: We went from a handful of autonomous robots that couldn't make it across a wide open desert to autonomous robots that are navigating real world urban traffic. Wow. THAT'S progress. Projects like those referenced in the thread are just masturbation, imho.

As far as humanoids even really being significant, I tend to disagree with the usual opinion. As soon as humanity crawled out of our collective caves, we invented the wheel. Once we had the wheel we invented the cart. And then the motorized cart. Why? Because our biological design is horribly inefficient for most work. Our brains allow us to optimize against our limitations - so now it's time to duplicate REAL biology: Combine brains with wheels (and their offshoots).

If walking was so efficient, we wouldn't have cars, trucks, motorcycles, bicycles, tanks, ... Henry Ford could have built a cam-drive walker - but he didn't. Go figger.

>> or simply envious of what they are doing?

I'm not picking on you, Layton - normally this would be a PM so I apologize for the public nature of this - but it seems like this topic may get heated and I'd like everyone to consider their response first.

Putting on my moderator hat for a moment: that whole statement was hazardously close to smelling like an Ad Hominem. If it was anyone other than Matt that was the target, it would probably be a problem. Let's please remember to attack the issue and not the person. Matt's robo-envy is not the issue : the issue is an opinion of whether or not we should approach areas of exploration in the way in which we are. Please focus on the issues rather than the person exploring the issues, if not for the sake of a person's feelings and the forwarding of open discussion consider how it lead already to a false assumption about a person by your own admission. And do we even KNOW if this is something being presented to us for the sake of sparking a discussion rather than Matt's strongly-held personal belief? I don't believe I know him well enough to make that call. Very few here probably do.

Thanks for reading!

Layton
09-04-2008, 05:21 PM
Hello.

Honnestly I don't think that these kinds of project specialy help for the interest in the general public. I find it very interesting/awesome that they can make a robot walk like this. But I talked to all my friends and family relatives. None of them knew ASIMO or other robots that were mentionned before in this thread.


Hm, well your little survey isn't going to accurately represent populations in Japan and South Korea, where robots like Asimo and Hubo are practically rockstars, commonly appearing in television segments and on-stage at concerts.

Personally I don't understand why ANYONE interested in robotics would be against any form of robotic development, period. And I also don't see how you can separate the cognitive functions from the body. The possibilities of a humanoid robot far outweigh those of a blackbox on wheels, and generally speaking stable walking technology is now not uncommon or difficult to do for these organizations, so that now the "brains" can be the focus.

I wonder if this bias against humanoid robotics is also felt towards the community who have invested in any of the kits sold on this very website?

Personally I'm more disgusted with all these "humanoid" robot projects. These projects generally don't generate lots of great vision/locomotion/etc research. It takes so much time to develop the legs and stability most projects run out of money before they get to the vision/speech/thinking/etc research. Why do we need legs? A wheeled robot with a human-shaped torso is every bit as humanoid as a walking one - or are you gonna argue that people in wheelchairs not as human as those who can walk? There are plenty of humanoid robots that don't have legs. Your argument is leveled at a minority of humanoid robots, and those are the ones that typically do have the funding to explore the vision/speech/thinking areas. Asimo, for example hasn't changed much outwardly in the past 3 years, because Honda has reached a point where they can focus on the innards.
Nearly everyone of these robots is based on some form of ZMP, which is inherently stable, but slow and useless in the real world. These walking bots are even worse than wheeled bots on such real world terrain as stairs and grass... Of course wheeled robots will be more useful in many out-door or all-terrain capacities but will people want a tank-treaded monstrosity roaming around their house or office? The types of applications assigned to humanoids like Asimo don't require him to be able to run around or perform acrobatic maneuvers like a soldier.
Walking bots also use more power, most of them are on a tether, or only run for 30 minutes. Battery technology is improving all the time, and will continue to do so as we move towards electric cars. The same ultra-efficient and cost-effective batteries that we will develop for our automobiles will be put to use in robots, don't you think? Besides, there was a time when computers filled entire labs -- just because something is too big and inefficient for practical use today doesn't mean that it will be that way tomorrow.
We've seen dozens of asimo-type robots over the years, and the only ones that do anything interesting are typically R/C controlled. Lets get some real intelligence and sensing first, then work on making it look cool.
Unfortunately, the same can be said of all robots - it's hardly a problem of one type of robot or another.

And as to business side, we've been saying robots to clean the house and do your chores are "just around the corner" for almost 50 years - give it a rest. These projects get the public hyped up - and then let them down... Even robots like the AIBO, whose dog form-factor inherentily lowers the expectations of the end user, is a dissappointment. Too much research is going into funky locomotion like walking, and too little into the important stuff - sensing/thinking.

-Fergs And scientists have been saying A.I. that can pass the turing test is right around the corner for 50 years, too. Does that mean we should stop working towards these goals, simply because we haven't achieved them? I'm afraid I don't understand your logic.

Adrenalynn
09-04-2008, 05:25 PM
>> The possibilities of a humanoid robot far outweigh those of a blackbox on wheels

Examples?

>> I wonder if this bias against humanoid robotics is also felt towards the community who have invested in any of the kits sold on this very website?

Issues, please.

JonHylands
09-04-2008, 05:42 PM
I'm going to throw my opinion in here. I believe we will initially achieve AI with a legged robot. The main reason is because most of the things that go on in our brains are just different versions of what it takes to walk (and no, I'm not talking about ZMP or IK).

Our brains are big predictive sequencing engines. That's what we do when we move, when we talk, when we listen, when we see things, and when we remember. Walking is the easiest of these things to do, and as soon as someone does it "right" I believe the other activities will be much closer to being solved.

Once we've bootstrapped an AI "brain", it won't matter what kind of body it is put into, but I think it will be a lot simpler to get there in the first place with a walking body.

- Jon

DresnerRobotics
09-04-2008, 05:43 PM
Layton I think you're missing the ultimate point. They said outright in the beginning they intended to market it eventually. I think Matt's point was that given our current level of technology and the current technical wall that robotics faces currently (ie: reliable navigation and location awareness, among other things) it's sort of silly for companies to be promising personal servants. It's one thing to do it for the sake of research, it's another when they're setting expectations that robots will be bringing you drinks in the near future.

Adrenalynn
09-04-2008, 05:44 PM
JonHylands>>I believe we will initially achieve

AI Real honest-to-goodness AI is being performed today in limited fashion on things that don't have ANY form of locomotion. Maybe you might clarify or define your "AI"?

JonHylands
09-04-2008, 05:48 PM
Real honest-to-goodness AI is being performed today in limited fashion on things that don't have ANY form of locomotion. Maybe you might clarify or define your "AI"?

I'm thinking "I Robot" (the movie) level of AI. An artificial being that can think for itself, than can do most of the things we can do.

Matt
09-04-2008, 05:48 PM
Let me clarify. I'm frowning on people thinking that sinking millions into developing a "household robot" as complicated as a humanoid as a product is in any way a good bet. I'm frowning on PR releases telling people this kind of technology is around the corner. I'm frowning on people believing that they could hope to make such a product in the next 5 years. I was making a point about how much such technology would cost versus buying the same "product" with current technology. IE: cheap labor versus a machine. Remember, new technology is only adopted by the masses when it's cheaper than the existing technology. All of this has nothing to do with harping on development, research, science, or types of robots. I'm merely talking about the productizing of technology. Bottom line, it's waaaaaaaaay to early to be talking about household humanoids as products. (in my opinion, which is just that, my opinion)

>>>>>>Personally I don't understand why ANYONE interested in robotics would be against any form of robotic development, period.

I specifically said that I agree with research being done. I'm all for development. I said that in my post.

>>>I wonder if this bias against humanoid robotics is also felt towards the community who have invested in any of the kits sold on this very website?

I can't follow that leap of logic there, sorry. I love humanoid robotics and so does this community. I wasn't bashing a type of robot, that would be silly. As Adrenalynn asked, argue and debate the ideas and the points. Attacking the person for not liking their opinion is not okay here.

Adrenalynn, feel free to lock the thread if you think I've kicked a hornets nest. I deserve a spanking probably! Sorry everyone, didn't mean to offend anyone, I was just going off on a silly rant. :P I'll be quiet now !

cheers,

Alex
09-04-2008, 05:53 PM
Personally I don't understand why ANYONE interested in robotics would be against any form of robotic development, period.I don't think anyone is saying that they are "against" humanoid robots, it just sounds like they are saying that legged robots are fundamentally flawed, and I tend to agree.

I cannot deny that humanoids and other legged robots are wicked awesome to look at and we, as humans, are preconditioned to be attracted towards any form of such. But, from a logical standpoint, in many applications, probably more so than not, legged robots just don't make much sense (edit: from a financial standpoint) to have in working environments. I'm sure this will change in time, but not for many, MANY years.

TechMech
09-04-2008, 06:01 PM
[deleted by Adrenalynn. Issues-focus, please.]

Layton
09-04-2008, 06:03 PM
>> The possibilities of a humanoid robot far outweigh those of a blackbox on wheels

Examples? What I meant by that is... take for example the roomba. It is only good at doing one thing. If however, you had a humanoid robot (even one on a wheeled base) assuming it has arms and hands, it could be programmed to operate a standard vacuum cleaner along with many other things. In other words, the possibilities are greater because it will be more adaptable to the environment we live in and the tools we use. Even if you have a robot which for all intents and purposes could not be described as a humanoid, it will still need at least an arm and hand to do anything useful, and my definition of a humanoid is broad enough to include such a robot.

Adrenalynn
09-04-2008, 06:04 PM
Au Contraire, mon Bossman...

This is a tough question, largely philosophical, imho. We have a tremendous amount of brainpower here on the TRC - why shouldn't we explore the tough questions?

I disagreed with Jon, and asked for clarification. He rationally clarified. My gut disagrees with his gut. I respect Jon and acknowledge I may be wrong. Heck, I have a counter-argument prepared for my "Henry Ford" statement. I have a counter to that counter too. We're not going to reach any kind of syllogistic answer here - there's no "truth table" when we're talking about "gut feelings"...

I'd rather not get all heavy-handed - let's argue like grown-ups [extracting the subset of politicians] and maybe we'll make a discovery. I'm certainly open to being "convinced".

sthmck
09-04-2008, 06:10 PM
Personally I don't understand why ANYONE interested in robotics would be against any form of robotic development, period. And I also don't see how you can separate the cognitive functions from the body. The possibilities of a humanoid robot far outweigh those of a blackbox on wheels, and generally speaking stable walking technology is now not uncommon or difficult to do for these organizations, so that now the "brains" can be the focus.

I wonder if this bias against humanoid robotics is also felt towards the community who have invested in any of the kits sold on this very website?

Hey,
I think that people can appreciate what these researchers are doing especially the amount of work that goes into these robots. I think that the real problem is that these projects just don't really do much. To be honest I think that comparable to size and budget the people building robo-one humanoids in japan have done just as much. The capabilities of these robots dont really justify the price tag. You can say well these robots have very advanced walking capabilities, but the truth of the matter is that you could just use two wheels.

Dont get me wrong the humanoids that are under development now do have impressive walking capabilities, but you never see demos of these robot out strolling through the park. A wheel works best on smooth surfaces, but atleast it can travel over some types of varying terrain. The only real advantage that legs have over wheels is the ability to move over uneven surfaces. When humanoids start doing this as fast as a four year old kid then we are getting somewhere.

I will say that I am a fan of the research being done by universities like MIT, CMU, and TU Delft that are working on dynamic walking. I think that humanoids in the future will probably use walking schemes develop on the bases of this research. The reason I say this is because those projects have proved to be much more energy efficient than the robots being made in Japan. Some even hover around the efficency of humans. I will admit that these robots really dont do anything but walk either, but the have been slated for market in the near future.

I will have to say that the robots that I am most impressed with are being developed at Anybots.
I dont know if anyone here has seen what they are doing but it is pretty cool. Yeah monty is teleoperated, but it still takes a lot of work to be able to get the amount of control that they are able to with it. And what they have done with dexter is equally as impressive. I doubt they have spent as much money on both of those robots combined as any humanoid being develop in Japan.

Blah :eek:.
So all in all projects like those humanoids are probably a big waste of money, and probably the money and research could be put to better use solving a more pressing problem. The fact of the matter is though if there are investors that are willing to spend the money than they will. So I will just keep plugging away on my projects in my basement.

Adrenalynn
09-04-2008, 06:15 PM
Ok, I see that. That's a step back from what I thought I was reading. I don't know that I would associate "Slap an arm on some wheels" with "it's a humanoid". I thought we were talking more about walkers/crawlers/shufflers?

I don't know that I'd diss the roomba-esque class (such as mops, vacuums, mowers, etal) that way personally. They are inexpensive task-focused devices. For me, that's like saying "I won't own a cell-phone because it can't brew coffee".

A hammer isn't a hex-head wrench. A hex-head wrench isn't a phillips screwdriver. A phillips screwdriver isn't a soldering iron. I still like my hammer, hex-head wrench, phillips screwdriver, and my soldering iron. Sometimes I use on or two of 'em whilst talking on my coffee maker - errr - cell-phone. ;)


What I meant by that is... take for example the roomba. It is only good at doing one thing. If however, you had a humanoid robot (even one on a wheeled base) assuming it has arms and hands, it could be programmed to operate a standard vacuum cleaner along with many other things. In other words, the possibilities are greater because it will be more adaptable to the environment we live in and the tools we use. Even if you have a robot which for all intents and purposes could not be described as a humanoid, it will still need at least an arm and hand to do anything useful, and my definition of a humanoid is broad enough to include such a robot.

Layton
09-04-2008, 06:34 PM
Ok, I see that. That's a step back from what I thought I was reading. I don't know that I would associate "Slap an arm on some wheels" with "it's a humanoid". I thought we were talking more about walkers/crawlers/shufflers?

I don't know that I'd diss the roomba-esque class (such as mops, vacuums, mowers, etal) that way personally. They are inexpensive task-focused devices. For me, that's like saying "I won't own a cell-phone because it can't brew coffee".
I'm not dissing the roomba, it's just limited compared to what a humanoid could do. But the cell-phone argument doesn't hold water. Cell-phones are more than just telephones: they're cameras, alarm clocks, notebooks, mini televisions, radios / mp3 players, etc all rolled into one. If anything, it proves my point that a universal humanoid would be a better value than 100 task-forced devices that can only do 1 thing. This is why people go nuts over things like the iPhone, because it does more for you than a rotary phone. Oh, and as for the wheels or legs debate, why not have the best of both worlds? Google Emiew-2 from Hitachi. Or go here: http://robotbling.gametrailers.com/gamepad/index.php?action=viewblog&id=324842

Adrenalynn
09-04-2008, 06:49 PM
I threw my cell phone away because it wouldn't solder and couldn't remove allen screws, so I'm not sure I'm qualfied to debate that. ;)

Seriously though: Humanoids as defined as legged robots (few humans have wheels at birth. Bipedal locomotion is a defining trait of "humanoids") can't do *anything* of value today. I haven't seen a single one that can walk without falling down. So assuming we CAN address that, and that we SHOULD address that - in some unknown number of years we'll have to approach the problems noted above re: navigation, localization, stair climbing, uneven terrain, load-handling, ... In the meantime, we have no walking semi-trucks but the wheeled ones (with EIGHTEEN of 'em, in fact) are hauling trillions of tons of usefullness - and there are very few people picking up 30,000 pounds to load those aforementioned semis without the help of a forklift or pallet-jack - neither one of which I've seen walking across the warehouse.

We don't call raccoons "humanoid", yet they have arms and "hands". Their "problem" is that they have too many legs to be humanoid... Sorry, but I think we're trying to redefine "humanoid" with the "wheeled arm".

American Heritage Dictionary says: "Having human characteristics or form" Unless I start seeing some humans born with wheels, I'm going to disagree with you on your extension of the definition.

Adrenalynn
09-04-2008, 06:56 PM
I'm thinking "I Robot" (the movie) level of AI. An artificial being that can think for itself, than can do most of the things we can do.

Sorry - I meant to reply to this:

Fair enough. I don't see this as an imperative tied to walking. In fact, I disagree completely: So much effort is given to not falling on its face in changing terrain, load, and environment that any AI developed is going to be "not falling on face focused"... ;) I don't see any required link betwixt AI development and a lack of wheels...

I appreciate your POV though - something to think through more re: will legs add weight to brains.

Thanks for the clarification!

TechMech
09-04-2008, 07:03 PM
I'm going log out and watch Transformers now .

JonHylands
09-04-2008, 07:05 PM
I don't see any required link betwixt AI development and a lack of wheels...

My point is not that its a requirement. It will just make the other (harder) things easier.

Adrenalynn
09-04-2008, 07:12 PM
My point is not that its a requirement. It will just make the other (harder) things easier.

Right, sorry, my misstatement. I don't see the association of making the other [harder or not] things "easier". I'm not making that connection.

Layton
09-04-2008, 07:16 PM
Robots on wheels can be semi-humanoid especially if they have two arms, two sets of hands, fingers, etc.. It may not fit into the same category as bipedal humanoids, but they are still more similar than a blackbox on wheels. Similarly a blackbox with legs is more humanoid than a blackbox on wheels (like the early Honda bots). See the following: Wakamaru (Mitsubishi), Enon (Fujitsu), PGR3 (Robotech), Emiew (Hitachi), Mechadroid Type C3 (Business Design Lab), Palette & Posy (Flower Robotics), etc etc etc I could go on but let's stop and look at Posy from Flower Robotics.

http://www.lovingthemachine.com/uploaded_images/Posy_s-774874.JPG

Are you telling me this robot isn't humanoid because it moves on wheels? Your narrow definition needs to be expanded to include more than just bipedal robots.

lnxfergy
09-04-2008, 07:22 PM
My point is not that its a requirement. It will just make the other (harder) things easier.

I can sort of see that. I recently was at a lecture by an Italian cognitive science researcher (who's name escapes me). His current research was based on the idea that our brains use internal models of how the world works to anticipate events, thus speeding up our response time. For instance, when trying to catch a ball we have an internal model of gravity that lets us decide where the ball will be...

His lab was working more with arms than legs, but his argument was that the same model-learning techniques used for catching a ball with an arm could be applied to speech and vision... It was of course really far off, but maybe....

-Fergs

Adrenalynn
09-04-2008, 07:27 PM
>> Your narrow definition needs to be expanded to include more than just bipedal robots.

Why? Can you please point-out humans born with wheels?

And is a robot with a couple salad bowls slapped on a neck-stalk more human-like than a wheeled robot with two arms? And why are two arms so hot? Why not four? Six? Sixteen? 394 and a tail? Clearly we have an opportunity to right the wrongs of biology, so why settle for two arms and inefficient legs when we could have 394 arms and a tail and roll around on six unstoppable tracks that function in any environment? And that whole head-thang with the forward-looking eyes - incredibly silly. I want eye-stalks with eyes that rotate 360 degrees. Scratch that, I want 360 eyes spaced 1deg apart. And I want them to have multiple facets capable of "seeing" into the microwave and down into VLF with everything in between.

Biology sucks. Legs suck. Forward looking eyes suck marginally less than side-facing eyes. Come on, we can FIX all this. The only thing we have worth writing home about is our brains, so let's plan on getting one or two of those working on this hardware and then we can build the uberbot...

metaform3d
09-04-2008, 07:31 PM
Wow, a proto flame-war in the TRC.

I've seen quite a few people catch the robot "bug" and start out with grand ambitions. "Why, in a few months I won't have to go to work any more..." These are sadly but ultimately subsequently crushed by the humiliating realities of present-day technology and the woeful lack of any workable theory of cognition. Household robot, especially humanoids, are limited by the three things that SciFi always gets wrong: brains, brawn, and batteries. All of those have to be addressed before we have Rosie, wheeled or not.

Wheeled robots will be as good as legged in a handicapped accessible environment. This means commercial and public buildings before personal homes.

Adrenalynn
09-04-2008, 07:38 PM
I'm in complete agreement with you with the exception of the "as good as [except in] personal homes". I've had the honor of playing with tracked robots that will blast around my two story home with reckless abandon. Stairs? Psh. 30mph up your funny little stairs. Fun toys (out of the hobby price range), that will run circles around any legged fall-ie thing, whilst carrying five hundred pounds around -- but they still lacked a brain. Closest they can come is "augmented control" - ie a computer helping speed-up the human's brain-motor inputs to keep up with the mechanical critter that is so fast and maneuverable that our wimpy little biology can't hope to keep up without some predictive help.

So I would say it's "brains, brains, brains, battery power, and brains".

JonHylands
09-04-2008, 07:43 PM
Right, sorry, my misstatement. I don't see the association of making the other [harder or not] things "easier". I'm not making that connection.

Well, lets look at each one. My premise is that walking is best solved using a predictive sequence engine. Of course, part of that has to include learning feedback, but fortunately that's not too hard to incorporate with a sequence engine.

Talking is just a different set of muscles (vocal cords), with sequences of pre-made sounds that we say. We use our own ears as feedback, to make what we say sound more like the word our parents were trying to teach us at the time.

Listening is another sequence, where you are doing prediction on what should come next, and walking a huge tree of possible remebered sequence chains to match up with what you are hearing.

Vision, and specifically object recognition, can be done in part using sequences. If you think about how graffiti (Palm) works, its a simple form of visual recognition, with a very small set of objects it recognizes. Each object is a sequence of strokes, which are scale independent (and could easily be rotation independent as well). Finding the outline of an object you are holding is easier if your eyes (cameras) can both point at it, so you can "lose" the background clutter. Trace the outline of the object, and you have something like a graffiti sequence, with the same scale and rotation independence. The real trick here is building a loose enough hash function so you can do a fast enough lookup on the sequence to find everything that is close.

Walking is the simplest form of sequencing, just because we (the developer) get instant feedback on the results, and its easy to keep track of the whole thing in your head while you're developing it.

Layton
09-04-2008, 07:51 PM
Can you please point-out humans born with wheels?



Interesting, you seem to suggest that a human without legs ceases to be a human. The robots I listed which you simply ignored have other human characteristics which must be considered regardless of whether or not their primary mode of transportation is wheels. And as for your second question, why not 16 or 394 and a tail, well if that is your idea of progress go for it, it sounds like the model of efficiency!

Let me know how that works out for you. Of course, be careful about putting your new incredible creation in a hospital or any other human habitat as it is likely to scare the patients to death or cause a panic in the streets as people flee in terror. Besides this problem, one other comes to mind: in your race to out-do mother nature you fell into the same trap as the humanoids you despise: creating something from the outside-in instead of starting with the brains, brains, brains.

Alex
09-04-2008, 08:00 PM
what is/isn't a humanoid is for another thread please. Actually... I think we already have one here started somewhere.

Adrenalynn
09-04-2008, 08:13 PM
in your race to out-do mother nature you fell into the same trap as the humanoids you despise: creating something from the outside-in instead of starting with the brains, brains, brains.

Please read my whole position statement:
"The only thing we have worth writing home about is our brains, so let's plan on getting one or two of those working on this hardware and then we can build the uberbot..."

Adrenalynn
09-04-2008, 08:18 PM
Walking is the simplest form of sequencing, just because we (the developer) get instant feedback on the results, and its easy to keep track of the whole thing in your head while you're developing it.

I respectfully disagree again. Walking is incredibly tough - hence the reason these multi-million-dollar 'bots like to faceplant when someone doesn't catch them first.

Contrarily, I'm selling face-recognition software today, I have sold video pattern recognition, and sold Iris recognition many years ago. I use speech recognition hundreds of times a day, and, well, you already have the Palm Graffiti example.

If walking were simpler than face/video/speech recognition, wouldn't I be using it on a daily basis like I'm using the other examples?

robot maker
09-04-2008, 08:18 PM
reading all these replys ,why do we make robots,just for fun or toys
i wounder about the first computer made was made only for fun and was going into production ,it took many years before anyone bought one and caught on and now everyone has a computer
i beleave the same with robot too,also heard that in less then 10 years they will have a robot car to drive people around,in japan they have some already and heard at the olympics they used robot walkers,also look at david hanson robotics they are really good ,AI ,FLUBBER closest design to real skin,expressions in his are the best i have seen,and his new design comming out at the end of 2008
ZENO named after his son
Our first incarnation is Zeno (http://www.zenosworld.com/)—a character robot that will walk and talk with kids and see them, all wirelessly controlled by a PC. Enlivened by Hanson’s famous conversational artificial intelligence software, Zeno will engage in complete dialogues with kids, see their face, make eye contact, and have his own moods and make his own decisions.
Zeno will be 17” tall, weigh 6 lbs. and run untethered on battery power. The body will walk and balance gracefully, and the face will use Hanson’s Frubber-based skin for best expressivity.
that from the website,and well under $500,heard $350,cant see how cheap it is i cant wait to buy one

robots i make are not just for fun,but a learning tool ,to make better
also i disagree with anyone saying saying robotics will not do house chores,i wounder also when the first robotic machine was made ,did they say never be used in production,now alot of car makers use them if not all
in medical i wounder the same question about replacing amputated legs and amps not the first robot arm was placed on a soldier,just a little while back and same has a real hand,and i thought i heard same was done with a leg too

DresnerRobotics
09-04-2008, 08:21 PM
reading all these replys ,why do we make robots,just for fun
i wounder about the first computer made was made only for fun and was going into production ,it took many years before anyone bought one and caught on and now everyone has a computer
i beleave the same with robot too,also heard that in less then 10 years they will have a robot car to drive people around,in japan they have some already and heard at the olympics they used robot walkers,also look at david hanson robotics they are really good ,AI ,FLUBBER closest design to real skin,expressions in his are the best i have seen,and his new design comming out at the end of 2008
ZENO named after his son
Our first incarnation is Zeno (http://www.zenosworld.com/)—a character robot that will walk and talk with kids and see them, all wirelessly controlled by a PC. Enlivened by Hanson’s famous conversational artificial intelligence software, Zeno will engage in complete dialogues with kids, see their face, make eye contact, and have his own moods and make his own decisions.
Zeno will be 17” tall, weigh 6 lbs. and run untethered on battery power. The body will walk and balance gracefully, and the face will use Hanson’s Frubber-based skin for best expressivity.
that from the website,and well under $500,heard $350,cant see how cheap it is i cant wait to buy one

robots i make are not just for fun,but a learning tool ,to make better

How will a robot built on the $1900 Manoi PF01 platform with a considerable amount of animatronics and AI software added to it amount to $500 or less?

Adrenalynn
09-04-2008, 08:22 PM
The first modern computers were built for targeting weapons systems - hardly "for fun" - unless sinking your opponents battleship was what you'd call "fun".

Ditto robotics - military and industrial. We "fun" types are johnny-come-lately.

Adrenalynn
09-04-2008, 08:28 PM
Our first incarnation is Zeno (http://www.zenosworld.com/)a character robot that will [...] sees, hears, talks and remembers who you are. [And otherwise knocks off the decades-old Teddy Ruxpin] [snicker]

Alex
09-04-2008, 08:35 PM
Hey Zeno rocks!! I saw him out at RoboDevelopment and I was really impressed. I even got to fondle that creepy flubber:D

robot maker
09-04-2008, 09:02 PM
here is a link from ZENO PRESS about $300 for the ZENO ROBOT
prices go down when in production,also saw links in many other places about selling for $300

http://zenosblog.wordpress.com/about/

http://www.zenosworld.com/



How will a robot built on the $1900 Manoi PF01 platform with a considerable amount of animatronics and AI software added to it amount to $500 or less?


maybe TRC will sell them soon


Hey Zeno rocks!! I saw him out at RoboDevelopment and I was really impressed. I even got to fondle that creepy flubber:D

sam
09-04-2008, 09:52 PM
Ok, that's it,

I leave you people alone here for a few minutes and I have a 5 page thread already?!

I was talking about a wheeled bot with arms. I didn't know that it was your definition of a humanoid. Again we can keep this for the other thread for this matter.

That kind of robot is pretty good for developpement. I think better than legged robots.

As for the AI system, it depends what AI is. My definition of AI will come in a kind of computer form for the military (of course). Jon, I don't quite get what you mean by I, robot smart. Saw the movie, maybe I don' reameneber correctly, but it seemed that they were VERY ineligent?

Sam

JonHylands
09-04-2008, 09:53 PM
I respectfully disagree again. Walking is incredibly tough - hence the reason these multi-million-dollar 'bots like to faceplant when someone doesn't catch them first.

Contrarily, I'm selling face-recognition software today, I have sold video pattern recognition, and sold Iris recognition many years ago. I use speech recognition hundreds of times a day, and, well, you already have the Palm Graffiti example.

If walking were simpler than face/video/speech recognition, wouldn't I be using it on a daily basis like I'm using the other examples?

The walking bots can't walk because they're all doing it wrong. We don't use ZMP, we don't use IK, we don't use math to walk.

You are using an implementation of those other things. But none of those things are integrated into a unified architecture. In other words, I think they are also doing it "wrong", from the perspective of how a person's brain does it.

This is all just my personal perspective, of course...

JonHylands
09-04-2008, 09:55 PM
As for the AI system, it depends what AI is. My definition of AI will come in a kind of computer form for the military (of course). Jon, I don't quite get what you mean by I, robot smart. Saw the movie, maybe I don' reameneber correctly, but it seemed that they were VERY ineligent?
Sam

Yes, the NS-5 was very intelligent, and that's what I'm talking about, what I'm aiming for.

robot maker
09-04-2008, 10:09 PM
JON
hope to see your idea work someday soon
thats something close to what i want to design for mine too,just takes time learning to me elctronics is easy,but just started learning programming and hope soon to be real good at it



Yes, the NS-5 was very intelligent, and that's what I'm talking about, what I'm aiming for.

JonHylands
09-04-2008, 10:23 PM
JON
hope to see your idea work someday soon
thats something close to what i want to design for mine too,just takes time learning to me elctronics is easy,but just started learning programming and hope soon to be real good at it

I'm exactly the opposite - electronics are difficult, but I was born with a keyboard in my hand.

sthmck
09-04-2008, 10:32 PM
Got to love the good old NS-5, which is probably the coolest humanoid robot ever in a movie imho. Did you ever notice how its actuators look like air muscles? lol

DresnerRobotics
09-05-2008, 12:14 AM
here is a link from ZENO PRESS about $300 for the ZENO ROBOT
prices go down when in production,also saw links in many other places about selling for $300

http://zenosblog.wordpress.com/about/

http://www.zenosworld.com/


I see someone in the comments section of a blog that hasn't been updated in 12 months talking about a $300 price tag. The initial models I saw looked to be using a Manoi body, the newer one doesn't appear to be quite the same.

That said- $300 won't even get you a Robophilo currently. I really doubt we'll see a robot with what appears to be $5000 worth of tech easily in the $300 range in a matter of a few years. I guess that makes two companies in this thread setting unrealistic expectations about their humanoids.

Matt
09-05-2008, 12:41 AM
My lord people. I go out for a few hours and come back to SIX pages, lol. I should be banned from posting in my own forums...

Let no man or woman claim we are not passionate about robots at TRC :) Many good points from all sides of the aisles here. Just remember, keep it as a spirited debate and no one has to try to "win" in here :)

carry on... :P

openmindedjjj
09-05-2008, 01:48 AM
i agree with matt, it seams whenever i post a form it gets all blown up and is poisoned with negativity,
i do agree with layton, shure humanoids are not that useful right now but so was everything else when it first started.( i mean look at the first car)if it wasent for japan and the other countries working hard to develop hunimoids we would probably not be able to buy a bioloid for 900 bucks or any other huminoid. give it 20 more years and just see how far they will have come. i mean 30 years ago people yousto think it was impossible to make a 2GB computer chip the size of a dime but now look.. i can buy a 2GB chip for my cell phone for 30 dollars. try to hold you toungs about how useless huminoids are for now. what is funny is probably half the people on here mostlikely own a huminoid or other simular robot.
[removed this line. - Adrenalynn]
i all most regret posting this considering the lack of excitement i was hopeing to see.
thankyou layton and anybody else who supported my excitement in the new robots which were being developed. and as far the leg vs. wheel design i think wheels are vary limiting i dont see how a wheeled robot could go up stairs or climb up a tree or anything other than drive on a somewhat flat surface.

Adrenalynn
09-05-2008, 02:39 AM
>> and as far the leg vs. wheel design i think wheels are vary limiting i dont see how a wheeled robot could go up stairs or climb up things or anything other than drive on a flat surface.

You need to look around a bit. Zero of the legged bots that I've seen can climb stairs. "Flame" goes on at-length about how cool it is that it can sometimes manage an 8cm drop without a faceplant, and a 4cm drop fairly consistantly. The stairs in my house tend to exceed 1.5-3.0 inches.

Hundreds if not thousands of wheeled (to include tracked) robots can climb stairs. You can buy stair climbers for almost nothing. SWAT relies heavily upon treaded stair climbers, as does the department of homeland security, and the US Military.

The rovers aren't legged bots either, yet they can explore the hostile terrain of mars.

Adrenalynn
09-05-2008, 02:43 AM
Well, I'm not sure our brains are doing it "all right" either. I'm yet to be convinced that the results of pseudorandom tampering combined with horrendously slow selection has yielded ideal results. Sure - it's yielded one application - the same way as our initial forrays have yielded applications. Are either of them "the best"? NP Complete. Neither of us can confidently answer that question truthfully. We can't answer that question without first inventing a "better way" - and then the answer to the question becomes self-evident. ;)


The walking bots can't walk because they're all doing it wrong. We don't use ZMP, we don't use IK, we don't use math to walk.

You are using an implementation of those other things. But none of those things are integrated into a unified architecture. In other words, I think they are also doing it "wrong", from the perspective of how a person's brain does it.

This is all just my personal perspective, of course...

openmindedjjj
09-05-2008, 02:50 AM
sorry what i ment was i dont think a wheel could compeat with a leg,
i mean a wheel can only spin "THATS IT"!!
a leg can run,jump,help climb,kick. dont get me wrong wheels can spin and they spin pretty fast thats why we use them for our cars, bikes, ect. but i think a leg can do more that a wheel. but thats is just my 21 year old opion. which in here is probably not much! lol :D

what they should do is put wheels on the end of arms and legs! like on transformers. that would be the best of both worlds.

Adrenalynn
09-05-2008, 02:54 AM
By that reasoning, legs lose again to wings, right?

darkback2
09-05-2008, 05:32 AM
Question...why make a humanoid robot to drive my car...when my car can drive itself? I used to visualize computer based brains that you could plug into your car, and in car mode it would drive you around...and then you could pull out that brain and put it into your vacuum and in vacuum mode it would clean your house...much like the NS-5...it is a human that could operate human like tools...but why bother...why not make all my tools smarter. Its sort of like in old houses that had a centralized motor system, and people used to connect hoses to do stuff...My aunts house for example has a motor/pump in the basement, and there are power takeoffs all around the house. Well...when the motor broke, all of the tools associated with it became useless. With a humanoid robot helper I still need to buy all of the appliances, and when the humanoid robot breaks down, then I have to do all the work myself. If I have all smart appliances then if one of them breaks then I can still use the rest. Computers are cheap. You can put them in anything.

But this is really two different arguments. Building an intelligent being is different than building something to do my work for me...

Predicting the future is silly though, because you can only predict it based on today. Maybe in the future we will have robotic bodies and all live in gell capsules...maybe that would a transition from what we are to something else...

Probably should just go to bed.

DB

JonHylands
09-05-2008, 07:52 AM
Well, I'm not sure our brains are doing it "all right" either. I'm yet to be convinced that the results of pseudorandom tampering combined with horrendously slow selection has yielded ideal results. Sure - it's yielded one application - the same way as our initial forrays have yielded applications. Are either of them "the best"? NP Complete. Neither of us can confidently answer that question truthfully. We can't answer that question without first inventing a "better way" - and then the answer to the question becomes self-evident. ;)

I have no expectation that the way I want to program my artificial brain is the "best" way. "Right" and "wrong" in this context simply means (a) is the technique scalable to where it needs to be, and (b) once it is set up, can it learn on its own (without more programming) to scale up?

ooops
09-05-2008, 08:57 AM
I get bizzy for one day, and miss all the fun.
Here is my 2 cents. If Honda, Toyota, Samsung and others are dropping millions of dollars on robots then guess what ... they know something we don't. I am not saying that their "marketing pitch" to the shareholders and the public that "house cleaning bots like "Rosie" are right around the corner" are true. They have to say that to justify the investments. Bear in mind these are the same companies that also spend millions on prototype autos that will never reach the masses. These are companies that are accustomed to showing you the very tip of the iceberg and keeping the rest under layers of security. Do I think they are focused on developing an army of "Rosies" no. I think they are in a race to develop military hardware more akin to Terminator than to Rosie. These companies may be full of it when they tell the public how close we are to Rosie, lets face it the militaries of the world will be willing to shell out big money on terminators long before you and I will shell out any money on Rosies. So, my opinion is if they have the money to spend on R&D great! If they tell us the public that it is for a Rosie to clean our house, take it with a grain of salt. As cool as robots are, these companies don't drop millions of dollars / pounds / yen over decades to just invent a one off marketing "toy". They have a goal in mind and they haven't shared their true intent, after all everyone loves Asimo, the public has a fear of technology to build a Terminator ... if you were the CEO of a billion dollar global company which one would you show the world?

openmindedjjj
09-05-2008, 12:44 PM
good point i had not thought of it that way.

Layton
09-05-2008, 01:26 PM
I get bizzy for one day, and miss all the fun.
Here is my 2 cents. If Honda, Toyota, Samsung and others are dropping millions of dollars on robots then guess what ... they know something we don't. I am not saying that their "marketing pitch" to the shareholders and the public that "house cleaning bots like "Rosie" are right around the corner" are true. They have to say that to justify the investments. Bear in mind these are the same companies that also spend millions on prototype autos that will never reach the masses. These are companies that are accustomed to showing you the very tip of the iceberg and keeping the rest under layers of security. Do I think they are focused on developing an army of "Rosies" no. I think they are in a race to develop military hardware more akin to Terminator than to Rosie. These companies may be full of it when they tell the public how close we are to Rosie, lets face it the militaries of the world will be willing to shell out big money on terminators long before you and I will shell out any money on Rosies. So, my opinion is if they have the money to spend on R&D great! If they tell us the public that it is for a Rosie to clean our house, take it with a grain of salt. As cool as robots are, these companies don't drop millions of dollars / pounds / yen over decades to just invent a one off marketing "toy". They have a goal in mind and they haven't shared their true intent, after all everyone loves Asimo, the public has a fear of technology to build a Terminator ... if you were the CEO of a billion dollar global company which one would you show the world?

The thing is, Honda's website explicitly states that Asimo technology will NEVER be used for any military application. And Toyota has outlined its intentions for the Partner Robot program already in August stating that by 2050 the Japanese population will be nearly half of what it is today. As the baby boomer generation gets older they will require more help at home and in hospitals and there simply isn't enough people to cover their needs. They plan on positioning their robots, including their mobility devices, as a means of prolonging independence at home.

The Japanese military is called the Japanese Self Defense Force for a reason. Since WW2 the army has been positioned as a means of self defense ONLY (outside of their UN obligations). The idea of that the military will use humanoid robots isn't feasible anyway; if you look at the military bots being developed for DARPA you'll see that they are primarily robotic humvees and the like. This is where the DARPA Grand Challenge comes in. They want to automate that stuff, not have terminator Asimos clumsily walking around and falling over at the slightest bump in the road. Humanoids like Asimo are only practical (if that) in a sterile environment where the ground is flat and obstacle free, and where they can create simple maps for navigation.

And no, Honda and Toyota are not saying their robots will be ready tomorrow or next year or in even 5 years, they're saying around 2020 or 2025. And when you look at the progress they've made in the last 5 years, I think that is reasonable. Even expensive robots that cost as much as a luxury car will make more sense than trying to hire an in-home nurse when none are available or stretched so thin they can't provide proper care, and this is what the companies are banking on. Imagine the situation in 20-30 years when baby boomers (now in their 60s) need 24h care!

for those who might be interested in seeing Toyota's current model:
http://i242.photobucket.com/albums/ff208/BaronCastle_2007/toyota_P1240830.jpg
http://i242.photobucket.com/albums/ff208/BaronCastle_2007/toyota_P1240824.jpg

metaform3d
09-06-2008, 01:05 AM
I'm with Layton on the Asimo. Honda has identified a huge market opportunity, and I mean huge. Automated eldercare is a fraction-of-a-trillion dollar industry, if -- and this is a big if -- the problem can be cracked. The patent licenses are worth god knows how much more to apply to other applications (*cough* military *cough*). They're a big company, they can afford to do the basic research. It might pay off huge, or it might all be lost. I'm OK with them taking a shot.

MarkosR
09-06-2008, 01:21 AM
Hi All,

I read half the resonses and wanted to add my 2 cents worth in two key areas discussed in this topic:

1) Why invest millions in flare robotics. I believe a good simile is when we look at the millions invested in Fromula 1 racing. These cars never will be owned by the avg person but it builds hype and prestige that ultimately funds projects by manufacturers. As one comment made by yall, it encourages interest and u create groopies like me who then indulge and invest

2) AI programming has many ways to be implemented. You can develop heuristic functions and develop learning trees to mimic the human brain. Storage will be the only constraint.

For me once battery technology evolves and we can use pico motherboards for longer battery life then we shall see some miracles. Until such time we need bigger servos to move the huge batteries :) Or stretch the limitations of the ATmel and the like controllers.

Slightly off topic I have found using Atmel and servo controllers managed remotely by PC's resolves the battery issue as well as the processing and memory constraints. Bluetooth is fine but too small a range. Use XBee pro or Xtend with 1 to 40mile range then ur cooking.

Thanks for lending me your ears hope ur not less intelligent for reading my ramblings..

metaform3d
09-06-2008, 03:05 AM
Hundreds if not thousands of wheeled (to include tracked) robots can climb stairs. You can buy stair climbers for almost nothing. SWAT relies heavily upon treaded stair climbers, as does the department of homeland security, and the US Military.droidcommander was interested in the specifics of stair-climbing robots on this thread (http://forums.trossenrobotics.com/showthread.php?t=2317). Low-COG tracked bots like iRobot's Packbot can certainly do it, but are there designs that work that don't look like an RC tank?

Environments intended for ordinary humans contain a lot of hazards for the wheel-bound, robot or otherwise. Steps, curbs, bumps and uneven terrain are common. Stairs, ladders and ledges with handrails are optimized for legged locomotion. Finally most items that a robot would need to manipulate are placed to be convenient for standing adults, 0.8 to 1.8 meters off the ground.

Given these constraints it's easy to see why employing a high GOG with some kind of dynamic balancing would be a better solution. Passive-balance systems have to keep the COG so low relative to the support polygon that it's going to be hard to interact with people that have their eyes and attention at around 60 inches high, while still having a base narrow enough for a normal doorway. I think the wheel is a great invention; I just have a hard time seeing how it can deal easily with these issues.

Handicapped accessible buildings mitigate a lot. Ramps where there are steps; elevators where there are stairs; wider doors; lower door handles and other affordances. I think building a general-purpose robot for that environment would remove some of the more challenging problems that will be encountered in the more typical home environment.

Adrenalynn
09-06-2008, 04:09 AM
I don't think that the whole COG has to be high to get into a human's site-line. Pedastals don't shift CoG much if at all, and certainly it moves it into our field of view.

What was your thought on "Wegs" and similar? I'm terribly fascinated by their design.

Stair climbing wheeled-like bots are a common college engineering problem. Some pretty cool designs have been implemented. My favorite looks like a big screw. It just twists its way over obstacles.

Regardless, at some point we have to look at the realities: Billions of dollars invested - and I've not seen a single viable humanoid stairclimber. But for a couple hundred bucks you and everyone else here can slap together a wheeled stair-climber. There's dreams and fantasies and there's what works [today]. Dreams and fantasies are fun, so I'm planning for my extended vacation on Calisto - but I still have to live and work here on Terra Firma.

A-Bot
09-06-2008, 09:58 AM
Dreams and fantasies are fun, so I'm planning for my extended vacation on Calisto - but I still have to live and work here on Terra Firma.

Io would be much more interesting. :p

asbrandsson
09-06-2008, 12:57 PM
I don't think anyone is saying that they are "against" humanoid robots, it just sounds like they are saying that legged robots are fundamentally flawed, and I tend to agree.

I cannot deny that humanoids and other legged robots are wicked awesome to look at and we, as humans, are preconditioned to be attracted towards any form of such. But, from a logical standpoint, in many applications, probably more so than not, legged robots just don't make much sense (edit: from a financial standpoint) to have in working environments. I'm sure this will change in time, but not for many, MANY years.

Hello,

I think that the future will be more and more towards humanoid robotics as people find it to easier to relate to AI that looks more like a human and less totally foreign.

I think that all the technology currently available makes production of a relatively inexpensive humanoid robot commercially available - except that all the best programs are very proprietary - just look at evolution robotics vision software. I think that this has lead to alot of the best research being done over and over again - instead of people focusing on really developing something new.

Just look at the high end robots that are predominantly sold to universities- the HOAP series off the top of my head. You are paying from 50,000 to 500,000 USD for a robot that you can develop at home for less money. In fact for a half million I can put a team together to build a robot as good as Sony or Honda has ever built. So instead of adding to a base of knowledge at lot of great talent is being spent on re-inventing the wheel.

Asbrandsson

Adrenalynn
09-06-2008, 01:04 PM
>> find it to easier to relate to AI that looks more like a human

You know, it's funny - I don't personally know anyone that feels that way. They all, without exception, find humanoids "creepy". A couple of them refuse to be in the same room with humanoids powered on.

I know that's only an n=20 with a not terribly random sample, but it's still literally 100% - so I find myself questioning that oft-repeated mantra.

Layton
09-06-2008, 05:41 PM
[Bunch of attacking people rather than point deleted - Adrenalynn]


In one experiment, researchers brought in the robot to take part in the children's dance sessions and found that the toddlers would spend longer in the room if the humanoid was among them. On average, toddlers would stay in the room for twice as long when the robot was around. Mr Tanaka says researchers are increasingly convinced that children consider the robot not a toy or a living human being but "something between the two", a difficult idea for adults to understand. To contrast with the biped, researchers also gave the toddlers a simple toy that looked like a robot but cannot move by itself. The toy, named Robby, was handled roughly and constantly shoved to the ground - behaviour the toddlers would not show to their beloved walking, dancing humanoid. Children initially stayed away from the biped out of caution but gradually warmed to it, hugging the robot and otherwise showing affection. Initially, the robot would often fall over due to inconsiderate treatment. But after one to two months, the children would help the robot get back to its feet. Within three months, the toddlers would never allow the robot to fall. "They are adapting themselves to the robot and empathising with it, although nobody teaches them to do so," Mr Tanaka said. full article: http://www.theaustralian.news.com.au/story/0,20867,18829729-23109,00.html
Biased personal anecdotes are fine, but I prefer facts myself.

Adrenalynn
09-06-2008, 08:17 PM
Well, then let's talk about the fathers of psychology and psychiatry instead of anecdotes and spin:

Theory supporting my observations: http://en.wikipedia.org/wiki/The_Uncanny

Excerpt:
The Uncanny (Ger. Das Unheimliche -- literally, "un-home-ly") is a Freudian (http://en.wikipedia.org/wiki/Freudian) concept of an instance where something can be familiar, yet foreign at the same time, resulting in a feeling of it being uncomfortably strange. [1] (http://en.wikipedia.org/wiki/The_Uncanny#cite_note-unheimliche-0)
Because the uncanny is familiar, yet strange, it often creates cognitive dissonance (http://en.wikipedia.org/wiki/Cognitive_dissonance) within the experiencing subject due to the paradoxical nature of being attracted to, yet repulsed by an object at the same time. This cognitive dissonance often leads to an outright rejection of the object, as one would rather reject than rationalize.

[...]

The state is first identified by Ernst Jentsch (http://en.wikipedia.org/w/index.php?title=Ernst_Jentsch&action=edit&redlink=1) in a 1906 (http://en.wikipedia.org/wiki/1906) essay, "On the Psychology of the Uncanny." Jentsch defines the uncanny as: "doubts whether an apparently animate being is really alive; or conversely, whether a lifeless object might be, in fact, animate" [1] (http://en.wikipedia.org/wiki/The_Uncanny#cite_note-unheimliche-0), [...]

leading to: http://en.wikipedia.org/wiki/Uncanny_Valley

Excerpt: The uncanny valley is a hypothesis that when robots (http://en.wikipedia.org/wiki/Robot) and other facsimiles of humans look and act almost like actual humans, it causes a response of revulsion among human observers. The "valley" in question is a dip in a proposed graph of the positivity of human reaction as a function of a robot's lifelikeness. It was introduced by Japanese (http://en.wikipedia.org/wiki/Japan) roboticist Masahiro Mori (http://en.wikipedia.org/wiki/Masahiro_Mori) in 1970 (http://en.wikipedia.org/wiki/1970), and has been linked to Ernst Jentsch (http://en.wikipedia.org/w/index.php?title=Ernst_Jentsch&action=edit&redlink=1)'s concept of "the uncanny (http://en.wikipedia.org/wiki/The_Uncanny)" identified in a 1906 (http://en.wikipedia.org/wiki/1906) essay, "On the Psychology of the Uncanny." Jentsch's conception is famously elaborated upon by Sigmund Freud (http://en.wikipedia.org/wiki/Sigmund_Freud) in a 1919 (http://en.wikipedia.org/wiki/1919) essay, simply entitled "The Uncanny" ("Das Unheimliche (http://en.wikipedia.org/wiki/Das_Unheimliche)").


Further, you can't run around accusing people of bias without at least some attempt to identify bias.

Your paragraph countering my observation appears to have some issues: Five year olds are more accepting of "the strange" and "the odd" because, at that age, everything is "strange and odd". They haven't even finished solidifying their deep neural pathways.

If your argument is that if we exposed all children to humanoids, they'd be considered "common place" in a couple hundred years following full immersion - I'd probably agree. However, I suspect you'd have to quell the concerns of billions of adults living on the planet first. Therefore, the introduction must be made carefully and slowly.

Layton
09-06-2008, 10:22 PM
Adrenalynn, you're once again foiled by your own argument. The famous uncanny valley is a dip in the chart occupied by androids like the Geminoids, or the creations of Hanson Robotics - robots that look almost human but not realistic enough to trick the eye (leaving the viewer with a sense of looking at an animated corpse), and does not apply to robots like QRIO and Asimo, robots that would never be confused for a human being.

I have read articles where scientists tested people's reaction to robots, ranging from the very mechanical to the very human (one of the androids they used was the Hanson Robotics' Philip Dick android in comparison to Mitsubishi's Wakamaru, filling in the gap by using various gradations between them by morphing one into the other). The subjects found the more mechanical humanoid (Wakamaru) to be the least disturbing of the group.

Androids like the ATR Geminoids and Actroids are in the minority of humanoids (and I have never seen them walking around), they mostly just sit or stand still and barely move their arms - hardly representative of the humanoids we are talking about in this thread (like the Samsung / KIST Mahru that started this off). In actuality they have more in common with animatronic special effects used in horror movies than robots like Asimo or the Toyota Partner Robots, and should probably be considered their own group.

But what is more to the point is that once we have conquered the uncanny valley and have humanoids that look so natural you COULD mistake them for a human being, you can be sure that people will feel more comfortable interacting with these AIs than any other form of robots, and may even form long-lasting and rewarding relationships with them based on conversation and even ''personality''.

Doctor Robotnik
09-07-2008, 03:11 AM
Interesting, you seem to suggest that a human without legs ceases to be a human. The robots I listed which you simply ignored have other human characteristics which must be considered regardless of whether or not their primary mode of transportation is wheels. And as for your second question, why not 16 or 394 and a tail, well if that is your idea of progress go for it, it sounds like the model of efficiency!

Let me know how that works out for you. Of course, be careful about putting your new incredible creation in a hospital or any other human habitat as it is likely to scare the patients to death or cause a panic in the streets as people flee in terror. Besides this problem, one other comes to mind: in your race to out-do mother nature you fell into the same trap as the humanoids you despise: creating something from the outside-in instead of starting with the brains, brains, brains.
It's not that the removal of legs that makes them not human. Humanoid is not equal to human. Humanoid is two arms, two legs, a body and a head. Without legs it is not humnaoid. I call robots that waist down do not have humanoid features a j5, from Johnny 5 of Short Circuit, Robots with a head, arms, but use wheels treads. ect for locomotion.

Adrenalynn
09-07-2008, 05:07 AM
Adrenalynn, you're once again foiled by your own argument. The famous uncanny valley is a dip in the chart occupied by androids like the Geminoids, or the creations of Hanson Robotics - robots that look almost human but not realistic enough to trick the eye (leaving the viewer with a sense of looking at an animated corpse), and does not apply to robots like QRIO and Asimo, robots that would never be confused for a human being.

Absolutely it applies. It's a non-linear function. As they approach familiarity but are not confused for humans, they become less appealing.



I have read articles where scientists tested people's reaction to robots, ranging from the very mechanical to the very human (one of the androids they used was the Hanson Robotics' Philip Dick android in comparison to Mitsubishi's Wakamaru, filling in the gap by using various gradations between them by morphing one into the other). The subjects found the more mechanical humanoid (Wakamaru) to be the least disturbing of the group.


Alas, you make sweeping statements and mention documentation, but fail to provide any. It's impossible for us to apply or rate such things. Either studies have been done using proper protocol and you should then post them, or they haven't. Without citing references, we can't know what the results were let alone what biased the studies.



But what is more to the point is that once we have conquered the uncanny valley and have humanoids that look so natural you COULD mistake them for a human being, you can be sure that people will feel more comfortable interacting with these AIs than any other form of robots, and may even form long-lasting and rewarding relationships with them based on conversation and even ''personality''.

Although that might be the case, I can't be "sure" of any such thing. There's insufficient data for such certainty. Science Fiction authors have frequently postulated otherwise in common literature - "Bladerunner" comes immediately to mind.

Today none of these things can stand still for any extended length of time without face-planting in my experience, let alone carrying out any critical tasks with reliability. I saw Asimo have to be caught twice by his "handlers" in Vegas a few years ago... Turning over child care to such a critter would be like turning over childcare to a drunken crackhead - I don't think CPS would be fond of the idea.

Doctor Robotnik
09-07-2008, 05:13 AM
I for one have no problems with a non humanoid robot. Maybe this applies more to those who normally have little interaction with them?

Adrenalynn
09-07-2008, 12:24 PM
Sure. Which would represent 99.9999999999999999999999999999999999999% of the world, right? :)

sam
09-08-2008, 03:55 PM
of course :rolleyes:

This is getting really interesting, good job people. Nice debate so far!