open source bipedal robot ideas

[Hank Fletcher]

I decided to start this thread in response to recent interests that have cropped up on the forum with respect to bipedal robots. Yes, they're wholly impractical (although compared to a dalek or R2-D2, I'm not so sure), but there's no question the idea itself is as cool as all get out. I'm no psychiatrist, but if I was, I could probably make a career out of what I would call "the Giuseppe complex." But knowing that there's probably something pathological about this pursuit isn't what you came here to read, so let's get cracking.

You're doing it all wrong. Presume you've been doing it wrong all along, and everyone else has got it wrong, also. If you can get yourself to that stage, you've already made great strides beyond what anyone else has ever accomplished. Presuming everyone has been wrong will lead you to the next stage in building a bipedal robot: innovation. But before you move on from stage two, return to stage one: you've got it all wrong, and here's why.

There are already bipedal machines, and they're called humans. If there is a Giuseppe complex, the first step towards coming to terms with the complex is recognizing that the desire to build a bipedal robot is largely psychologically, and not utility, driven. You want to build a bipedal robot because you want to make something more human. This might seem like a digression, but it's not. In fact, it's a very practical step in making a bipedal robot. You want to make a bipedal robot because it's something you must do because you love it, and to that end it is a utilitarian venture, since anything else in life is merely service to your passion.

Knowing that the desire to build a bipedal robot is largely a psychological condition, in what way does that define the nature of your objective? This is important, because if you can define your objective, if you can articulate as best you can your goal, then you're a long way towards getting yourself from Point A (where you are now) to Point B (going jogging with C3P0).

So everyone's been wrong, and it's psychological. What does that mean to you? Or more fairly, what the heck am I talking about? To better explain where I'm coming from, let's talk about the RoboCup. If you don't know what that is, check it out:
https://www.robocup.org/
The idea in a nutshell is, well, it's right on the homepage:

By the year 2050,
develop a team of fully autonomous humanoid robots that can win against the human world soccer champion team.

If there's one aspect of this amazing endeavour the organizers have failed to fully address, it's the human element. This is, in part, where the psychology really comes into play. Imagine you're a human soccer player: wouldn't you think it reasonable for you to put certain conditions in place about what the robots can be? Do you really want to share the field with ED209? A sliding tackle would be the least of your worries.

The goal of the 2050 cup is a great incentive for young robot builders. Think of it: you could be the designer, owner, and coach of the best team in the most popular sport on the planet. But think what the limits might be from the humans: presume your robot will have to be about 6', 200lbs. Think about the challenges: your robot will have to be fast, flexible, damage-resistant, water-proof. Seems like a big challenge, right? But the goal is still a long way off. Psychologists say it generally takes about ten years of devotion to any one field to become an expert in that field. Multiply that by four, and that's where you'll be by 2050 - the top of one of the most lucrative athletic-entertainment activities in the world.

But how can you get started? Where do I get the nerve to say you and everyone else have been wrong so far? How can you get a leg-up on the competition today? My answer, as glib as it may seem at first, is for you to read this:
https://en.wikipedia.org/wiki/Walking
and come back with your ideas.

It's gonna happen, and some people are gonna make a lot of money from it, or at least have a heck of a lot of fun trying. Will that be you?

 

[HarveyH42]

Quite an ambitious venture... Last humaniod robot I saw was kind of slow and clumsy, it was tethered to power and processing. I don't see it running or kicking a ball anytime soon, but that was years ago. Think it was called HONDA, but that might have been the company financing the research.

Second point, is saftey. Machines don't think, feel, or even care if their actions cause injury or damage. The robo-soccer team would need extensive testing to prove safe for human interaction. Not really into soccer, but not technically a 'contact' sport, there are often serious injuries. No world class team, champion or otherwise would take that risk.

Computer technology will need to many huge advances in speed and computational power. It's a very fast paced game, to keep up with a human...

I think just getting a team of robots together, that could actually play the game would be a huge task, beating even grade school kids would be much more difficult.

 

[kjennejohn]

OK, for the general edification of myself: who is Giuseppe and what's his complex; and what/who is ED209?

Interesting idea though: a bunch of creatures made from metal, hard plastic and glass ramming and sideswiping humans in a competition aired around the world to screaming, beer/wine/saki-swilling bloodthirsty fans. I take it the human team of 2050 is composed of condemned prisoners? Capital punishment with a scoreboard. The game starts with the hosting country's national anthem and last rites. Red-flagged robots are sent to the crusher. RoboWars pales in comparison.
What I'd like to see is two teams of bipedal robots with individual intelligence meet in a match. C3PO with attitude.

Actually, I can see a need for bipedal robots with human-like hands. We can move these into industry and repair jobs without retooling the entire infrastructure and leaving humans unable to do anything. Robots would be able to move about any building designed for human occupation. I can't wait to find myself in a care home with English-speaking attendants with hands colder than the bed pan they're shoving under my butt.

Thanks for the Wikipedia reference. Very interesting.
kenjj

 

[HarveyH42]

ED209 is a movie robot (Robocop), had a few glitches...

 

[Nigel Goodwin]

ED209 is a movie robot (Robocop), had a few glitches...

Nah! - he performed wonderfully!

 

[Hank Fletcher]

Thanks for the Wikipedia reference. Very interesting.
kenjj

I hope you're being sincere, because I certainly was. The part in the wiki reference on the biomechanics of walking was the thing that really inspired me to start this thread. As you can tell from the comments on this thread, and the servo-operated bipedal robots some folks have tried out so far, there's something broken in our thinking.

My opinion is that we need to start thinking more about the mechanics of walking, and how we can conform to humans in a more realistic manner. We need to stop thinking about clumping, shuffling, or rolling along. We need to start thinking about walking more as a balancing act on one foot, alternating between two feet.

By the way, Giuseppe was an old man with no family, who made a puppet for his son, called Pinocchio. If I remember the story, his first prototype had a few glitches.

 

[kjennejohn]

Nah! - he performed wonderfully!

Is that the robot with a head like a Norelco razor, the stair-challenged model? Ah, yes, I'd hate to be the student in a school bus HE caught speeding... ;>)

Actually, why limit the robot to being bipedal? Aren't we handicapping him? Let's create a locomotion system that gives him/her/it much more mobility and includes walking and stair-climbing as a subset. Give it quadraped capability, so it can carry stretchers on either side; clamber about boulders and cliff faces like a mountain goat for search and rescue; carry heavy loads over thin ice because the weight is spread over four surfaces instead of just two. Think about adding a mechanism to the robot that engages a rail on the outside of buildings so it can run up and down like an elevator, at will. But leave two of these appendages hand-like so he can stand upright and use tools and machinery designed for human operation. There may be a psychological comfort in working with something near-human, but would you be intimidated working with a creature that looked like Mr. Ed that could make sandwiches for your lunch before plugging itself in for its nightly recharge?

I do so love the sound of my own typing ;P
kenjj

 

[blueroomelectronics]

ED209. I think it's a decent design.
**broken link removed**

 

[Hank Fletcher]

Actually, why limit the robot to being bipedal?

Because that's what this thread is about. I know quadrupeds and other creatures with more or less legs have their merits, but if you want to talk about those, start your own thread.

ED209. I think it's a decent design.

It looks cool, but it's only a decent design in the same terms as the fellows who do interior "design" on HGTV pick paint and furniture (incidentally, my girlfriend was on an episode of "The Designer Guys"). There's a difference between that type of design and the engineering of home building. There's a difference between sci-fi special effects and real-world engineering. There's a difference between fueling your imagination, and having your imagination fool you.

I think an analogy for comparing ED209 to a realistic starting point for a bipedal robot would be for someone to say, "Hey, we need a new design for a toaster," and someone else saying, "Did you see that movie Star Wars? They have this thing called a light-sabre. It slices and burns all in one!" So then the R&D department is told by management to make a light-sabre model toaster. Is it the most practical means of slicing and burning bread? Is it even a possible means of slicing and burning bread? So why would you use that as your starting point?

Don't get me wrong, I'm all for imagination, and in some situations, the more fantastic the better. But the reason I emphasized the importance of articulating the goal earlier is because the goal will largely dictate your starting point. Is the goal to have something that looks cool (which there are plenty of already) or to have something that works well (which is more rare)?

 

[Hank Fletcher]

Here's a video of Germany vs. Japan in the humanoid division of the RoboCup.
https://www.humanoidsoccer.org/media.html

Did you note the mechanics of their movement compared to the mechanics of their kicking? It seems to me that there are four basic kinds of movement:
1. The lumbering walk;
2. The sideways shuffle;
3. The stationary kick;
4. The flop.

What I mean by the "lumbering" walk is that the robots move forward with no part of their mechanics design towards the pendulus motion described in the wiki article. They merely lift one leg up, move it forward, and fall on it. There's no upper-body motion, with the exception of the little robot moving his arms. I think moving the arms is a step in the right direction, but the importance of shifting the upper-body in empathy with the legs is what I'm getting at. If you read the wiki article, you probably appreciate the value of the pendulum in terms of added speed and efficiency.

The sideways shuffle is something I suppose humans do. I guess when we do it, the pendulum aspect doesn't really come into play. Maybe the robots have this part more closely representative of human motion, but I don't know for sure.

The stationary kick in comparison to the lumbering walk demonstrates that the robots in the video at least have the capacity to balance on one leg, if not in a continuous, pendulus motion. In general terms of creating a bipedal robot, I think the stationary kick should be recognized as a potential link between the lumbering walk and the more ideal pendulus walk.

I don't know much about hockey, but I know enough to know that the flop is by no means a goal-keeping strategy to be proud of. I'm pretty sure it's probably the same for soccer, too. Leaning and falling so your robot blocks the ball's path will get the job done, but the ultimate bipedal robot will be more capable in human terms. It should be able to jump, and be able to direct its jump to any of 360 degrees from its orientation. I think this is going to be one of the final obstacles to overcome in the design of soccer-playing robots, but having the strength to jump is important. The same strength is needed to evolve the bipedal robot from a walking robot to a running robot.

 

[blueroomelectronics]

It's the chicken leg design I like. For something that doesn't need to sit in a chair the reverse knees may offer an advantage.
**broken link removed**

 

[Hank Fletcher]

It's the chicken leg design I like.

I've never really been aware of the virtues of that design, even as it exists in nature. Is there a more formal term for it, so I can read up on it more? It seems to be a design that favours using legs more for standing than moving. That seems to make sense considering chickens use to use their wings more for movement than their legs. Maybe chickens will evolve to forward knees someday?

I can't quite get my head around it, trying to compare the physics of both knees. Considering the pendulum motion as the goal, does the physics of either knee add up to the same thing? That is, does the reverse knee negate the efficiency of the pendulum? I'd tend to think not - ducks waddle, i.e. move their upper body as they walk. But then, they're better flyers and swimmers than they are walkers.

I couldn't find much on wiki about non-human legs and knees.

 

[blueroomelectronics]

Most animals hind legs are backwards knee, I think it's for pushing forward. Look at the ostrich, it doesn't fly but it can move and those legs pack a punch.

If humans evolved from ape (made for climbing not running) then maybe we're not the best design for running (sitting is good though)

PS it's not just the knee, the foot and ankle are also quite different.

 

[Hank Fletcher]

Good point, especially about the ostrich. I guess as far as natural bipedal designs go, it's probably got humans beat for speed, but can it jump? I know this is a bipedal thread, but how will an ostrich throw-in the ball at the RoboCup?

But man, those suckers can move!
https://www.youtube.com/watch?v=8Vgkn_gWBwg

 

[blueroomelectronics]

Doesn't have to jump if it's at the net before any forward knee design.

 

[Hank Fletcher]

I wish I could find more video or info about the mechanics of the ostrich legs. They don't seem to be even comparable to what humans have in terms of design.

https://www.youtube.com/watch?v=rhpVgUCHDIE&mode=related&search=
Watching them run, the ostrich's body seems to remain almost completely stationary in relation to their legs. They're going faster than humans, with no pendulus motion, as far as I can tell. I wonder what that means in terms of efficiency? You can see the neck wiggle, though.

Here's another fast one:
https://www.youtube.com/watch?v=7r_B-CM79-M&mode=related&search=

This is what the robot should be able to do after scoring at the RoboCup:
https://www.youtube.com/watch?v=2zQ-eVqX_mY&mode=related&search=

 

[blueroomelectronics]

I know it's scifi but the kid at the end of the movie "The Arrival"
<spoiler warning>



morphed into a reverse knee kid and ran off like the roadrunner.

 

[Hank Fletcher]

I'm intrigued, yet refuse to consciously watch any film I know contains Charlie Sheen (with the exception of perhaps Being John Malkovich). Denise Richards, on the other hand...

 

[kjennejohn]

Blueroom, those "chicken legs" are just the coolest thing. Can you provide a link to the entire design or give a picture of the whole creation? I would think that the more crouched the stance the better, to a point. That thought makes me appreciate just how tricky our stance is, our legs are mostly straight up and down. I have to wonder what advantage that gives us.

Eventually you have to consider how to recreate the muscle action. Perhaps "muscle wire" will advance in the future to the point where you can actually perform muscle action as we have it now. At present it is too inefficient. It takes a lot of current and is slow to recover. Embedding a cooling system of some kind would speed up the last problem, but that adds to the complexity of the design.

No doubt bipedal locomotion will be one of the Holy Grails of Mechatronics.
kenjj

 

[Krumlink]

The backwards knee design is much more stable, seeing at the weight would be centered over the feet, rather than humans where we are prone to falling on our asses. They would fall on their face, but with a large feet that wont be a problem.

Kettering Kickoff went great, we won 2nd place and I am in the newspaper!