Is anyone working in foot and legs?

[Roberto Alcantara]

Hi guys,

I have many years experience in embedded systems and I will try to help project in this field too, but for now I would like to improve my mechanicals skills.

So, have someone  working in foot/legs projects to Inmoov for now? This is one thing I will like to do.

best regards,

 - roberto

[Amby Ambroise]

oui je woud like inmoov with foot no wheel please

[Roberto Alcantara]

ok.  I never thought about wheels except now ;)

[peter heim]

Holomonic or omni drive would be easier to do then legs trying to build legs that can hold up 20kg of robot would cost a lot of money
wheels on the other hand can be as simple as diff drive which can be built for around $100 or a full powered castor model for $1000
the other thing with wheels they work with off the shelf programs (ROS) so all the hard work has been done.
Holomonic drives can move through a house just as good as legs

[Amby Ambroise]

m i am working to build some leg for my robot

voir cture but with wheel like this can be cool but i prefere leg :-)

[gael langevin]

My initial plan was to create a sort of self balancing wheel locomotion, I have found some interesting projects which seem to be pretty good. The plan of course is to work with Arduino through MRL again. It seems faisible by a DIYer. I also really like the idea of self equilibrium and speed movements about that.
Now of course creating legs is very tempting, but the challenge is tremendous compared to making two arms move around.
It could result in never walking at all because of too many difficulties.
It seems like engineers have been working on this problem for their robots since many years already.

So I'm thinking of using both plans in one. The idea would be to set motorized wheels on the knees. This way when InMoov gets on his knees, he would be much lower and could drive around like as a self balancing robot with faster locomotion. (What a dream!)
Then getting back on his legs he could walk, and maybe climb stairs. (Another dream!)

[nomas...@gmail.com]

Self balancing? Sort of like a Segway?

[gael langevin]

@nomas,
yes, I just didn't want to use the brand name.

[gael langevin]

@Roberto Alcantara,
Any idea how we can implement the equilibrium of the whole body on the legs. I looked a bit, and I guess I should use a Accelerometer: MMA8453Q (~$2) and Gyroscope: LPY450AL ($4). but how would I organize the servos to react to them through Arduino and MRL. Please give me very simple answers, if you want me to understand keep in mind I'm a newbie and no electronician.
I found this article. Of course it is about wheels but I think the theory could be applied to servos...

[nomas...@gmail.com]

Hey Guys, check out this robot, which seems to be the most advanced humanoid yet built: http://spectrum.ieee.org/automaton/robotics/humanoids/kenshiro-robot-gets-new-muscles-and-bones  The muscles are a really cool design... 

[Roberto Alcantara]

Hi Gael,

Yes, I have a few ideias in my mind. To be honest, biped motion is my main interest in robotics. I started to look designs and found inmoov because I need the hardware to test this ideias.

My first try was this   http://ostatic.com/blog/a-new-open-source-robot-wows-with-its-soccer-skills     but these guys are using really expensive motors (very nice by the way) and I would like to produce something educational and more accessible. So, I forgot this option and started my own design, because I want to improve my mechanical design skills too.

You right the main sensors are both accelerometer and gyroscopes (3 axis each). But this is the easy part of the game. With at least 6 motors per leg  the control isn't easy to combine. Maybe we need something more powerful than arduino to compute (DSP) if we are talking about try to walk (to decide what best set of motion options and do not fall - we need a lot of math here ;-)  ).

Basically we need to react the sensor changes and give to cpu the math they need to compute best option to motor compensation. We have a few options here and my first try should be using neural networks.

I'm working in foot/legs design, you can see a foot preview here:  http://imageshack.us/photo/my-images/203/lateral1n.jpg/  .  For now i'm using step motors instead servos because I need force and precise control to first prototype.It is smaller then inmoov will need, I guess (somebody talk about 20kg for now, is it right?). 

[nomas...@gmail.com]

Here is a Hi Res picture of the above robot featuring the Giant Dynamixel servos in the legs: http://www.ais.uni-bonn.de/nimbro/OP/images/NimbRo-OP_1.jpg  Very nice functional design.

[Niels vd spek]

Hello Gael and all others.

This is my first post here, i love the work you have done so far on inmoov Gael it look very wonderful. I have always wanted to build my own 2 legs robot. But my skills are not that good. Also i love to build your robot as well, when i find a way to build or buy a 3d printer etc.  I hope to see how it is going.

Looking for robots i just found the progress of an American School for robotics Robot named: CHARLI-2 he can already dance the 'Gangnam Style'.

Those legs looks very easy to rebuild our selfs.  Please note that the Charli-1 does have an other leg structure.  http://www.romela.org/main/CHARLI:_Cognitive_Humanoid_Autonomous_Robot_with_Learning_Intelligence . 

Self i just brought me a set of makerbean's to see how var i can come, myself. I will use the .netmf platforms from GHI electronics as controlling system. It is much faster and has more options then a normal Arduino.  

[nomas...@gmail.com]

Here is a paper on how Charlie walks, going into great detail: http://scholar.lib.vt.edu/theses/available/etd-05042012-155706/unrestricted/Han_Jeakweon_D_2012.pdf

[Roberto Alcantara]

Nice paper, thank you. I'm adding in my library to read it fully.

[gael langevin]

Nice paper with loads of details and a few pages of math ;)

[Enrico Colautti]

How about Stellaris launchpad? Faster than Arduino and still small and simple to build upon.
http://www.ti.com/ww/en/launchpad/stellaris_head.html?DCMP=stellaris-launchpad&HQS=stellaris-launchpad-b
80MHz 32-bit ARM-4 processor...

P.S. I've been so impressed by this project that I started building my 3D printer to print this robot.

[hansf...@gmail.com]

Yea I am going to try out the Arduino Due. This board now has a 32 bit processor so I believe it may be able to perform all the calculations.
I think when using 2 of them and putting it through a wireless connection maybe ideal.

I am still in the process of getting some of my other items. Still looking for the cameras and power supply to run all the servos. Don't get me wrong I like the battery but I believe for all the development during the early days a power supply will be less headaches during charging and waiting.

Hans

[B Stott]

Hi,

I've been too overwhelmed to post with the lot of you all having taken to on InMoov like fleas to a cat. But, you are talking of balancing and electronic components. I had to add to your discussion

There is a great bargain for the STM32F3 Discovery board. The board is an MCU development board by chip maker ST.

Some kickers are:

The board is ~$10.00-$15.00 US
The mcu is more powerful than Arduino. Due too, if I remember right.
The mcu is ARM Cortex M3
The board is a 7 DOF IMU. This has: 3 axis gyro, 3 axis accelerometer, and magnetometer. It even does temperature but, no one talks about that.

So, balancing on the cheap. I do not know regarding software porting but, I did read someone here is very good at embedded systems?

I bought a STM32F3 board but, am mired with customizing a pile of servos, waiting for new a gear set (oops!) and wire and connectors. Busy trimming and assembling a pile of printed parts, and reading about MRL while playing with a camera. This kinda uses up any free cycle to play with the IMU. Funny.....

Oh, to those without a 3D printer. Stop thinking! Build, buy, borrow (hacker spaces.). Make it out of wood, legos, rods, old signs, cardboard and glue, anything.... If you are here - You must have a 3D printer. 3D printing for you is now like breathing without lungs! LOL...

Hey, Gael! :-)

[gael langevin]

Hey Brian:)
Nice to see you on the forum. To all the new builders I present you Brian! Brian has been around the project since I designed the hand, he also intended the Pittsburgh Makerfaire where he presented InMoov among other interesting things and printers.
If the forarms have easy removable covers , it's because of his thoughtful suggestion,
Brian has been printing InMoov since the very beginning and if I have brought it to this point, it is partly because of his powerful spiritual support.
His work can be seen here.
Thanks Brian.

[Alain Lowet]

Hello Guys,

I think a good solution will be a mix between wheel and feet, i think of a wheel that can be raised down along the legs and behaving with a self balanced control system for moving , but keeping the legs/foot system for other postures such as sitting, or walking for short distance, climbing, etc..

then the wheel will move up beside the leg.

[im8act]

Well, given that other, comparatively high-profile, well funded groups haven't been able to successfully develop and/or implement particularly effective bipedal systems, I wouldn't hold by breath on the ad-hoc inMoov community... Slow but steady development is often nature of open-source projects...

I will however mention the feasibility of ballbots. Basically, instead of wheels or legs, you balance your robot atop a ball, controlling itself through a series of servos representing the X, Y, and Z axises. This allows for more natural movement compared to wheeled robots and simpler implementation when compared to bipedal alternatives. Several implementations have been demonstrated, a few of which even use Arduino, and there's at least one with arms! If anything, a ballbot route would hopefully suffice as a short term, get-it-moving solution until properly functioning legs can be developed...

Best of luck!

[peter heim]

A ball robot is just as hard to use as legs when the power goes off the robot falls over and breaks, a holomonic base is stable and can move the same as a ball drive

but when the power goes off it just stops

I have had a inmoov robot on a holomonic base and it moves great the ability to translate sideways is very handy

[Tonerwerk Toner, Druckerpatronen kaufen]

Hello Peter!

and what kind of holomonic base do you have for the Inmoov?

Also kurz auf deutsch, ich interessiere mich für ein fahrbares Untergestell für den Inmoov und würde auch lieber eins auf Rädern basierte Variante bevorzugen, da ein Segwheel oder the Bot-Ball doch sehr gefährlich sind beim sturz.

Was verwendest du?


Boris

[peter heim]

Hi Boris

here is a link to the first holomonic base with a very early version of inmoov(my take).

http://escaliente-robotics.blogspot.com.au/2013/03/robbie-get-head.html

it used 8 motors 4 to steer the castors and 4 drive motors

I stopped work on the holomonic base because I needed the motors for the arms

and I could achieve the same control accuracy with a well tuned diff drive base for less effort

peter 

[Amby Ambroise]

hello i working to build my legs with dc motor you can see my work

Le samedi 26 janvier 2013 17:52:52 UTC+1, Roberto Alcantara a écrit :

Hi guys,

I have many years experience in embedded systems and I will try to help project in this field too, but for now I would like to improve my mechanicals skills.

So, have someone  working in foot/legs projects to Inmoov for now? This is one thing I will like to do.

best regards,

 - roberto

[Amby Ambroise]

[Peter Montgomery]

Hi Peter, 

I was taking a look at the picture that was posted and noticed that you were using what looks like linear actuators, is that the case? if so what kind? I would like to do that with my build as well. 

Thank you, 

Peter  

[peter heim]

Hi peter

They are trim tab actuators from a cessna aircraft i had laying around they have a limited travel and move slowly

thingiverse has a few actuator designs that are very good using a threaded rod. I have another proto type 

using a lead screw but it offers no real improvement over the Inmoov actuator

peter