Need Advise I Want To Start Building My First Humanoid Robot

[wgn_white]

Hello, first of all my name is Wiguan and please accept my apology for my noob questions and behavior.

This is my first post here.

I found this forum by searching at Google for a keyword "robot forum" so I assume this must be a great forum to ask around how to build a robot.  

I want to build a humanoid robot which can perform at least a movement like in the vides below.

[youtube]isnGkUcv30o[/youtube]

By doing some research.

I found the creator website, http://dancingdoll-rz.com/main.htm

It's in Japanese and I can't read those.

By browsing through his website, I make a wild assumption but I bet these images will help.









I never build a robot before, nor playing with any electronic kits.

But I have these skills which build my confidence to even think to start the project.

1. I can work with Aluminium to build the robot body.
2. I have basic electronics knowledge.
3. I know how to solder.
4. I know how to program in C++

But as a total newbie, I wish to know what is that micro controller that this guy use for his robot girl?
It's called Dollfie by the way.

Or does it a BEAM robot?

Or is it a combination of both?

But what I really wanted to build is a robot with smooth movement like the robot in the videos.
The articulated part (what we call it here?) are operated by servos or solenoids.

The robot should seems "intelligent" enough to operate, walk, avoid obstacle or like the Femisapien.
The robot movement also can be programable.
The robot also can be controlled using a remote control.
The robot also can be program to do certain movements for each button on the remote control.

Again I apology if this is too much.

Please point me a reading resources as well, I just didn't know what should I research for to achieve my goal.
(I've read the Reading Resources for Newbeginner in this forum, all of them.   )

Wiguan

[photomark]

Wow nice looking robot photos

In all honesty I would say that this is way out of your skill set .

Walking robots require some very precise engineering and machining not to mention a lot of computer power or even better a lot of computers.


However I don't mean to discourage you but some recommended reading would be ,Physics (university level ) Mathematics (differential calculus ) engineering mechanics and engineering computing , there are a lot of books around on mechatronics, you should also study up a bit on human anatomy and learn the mechanics of walking

This is just a start as you will also need a metal lathe and mill and some fabrication skills to go with them .

I only know basic C++ my self so I am not to sure if that will have the scope for a walking robot

The photos you show look like some ones final year university project in mechatronics .

Personally I would recommend you start with an easier project than a bipedal and get your skills up a bit 

[Fr0stAngel]

The most basic problem with humanoid robots is balancing... ( and ofcourse the fact that they cost too much $$ than some other types of robots)...
so basically , i would say that if you dont have any experience in robotics, dont go for a humanoid just yet (especially if you are doing t all by yourself)....start with a much simpler and basic robot with a differential drive etc....then when you are used to handling servos, and a little complex programming, then you can go for the complex robots..
but please be patient , as learning things takes time...
BEST of LUCK..!!

[wgn_white]

Wow nice looking robot photos

In all honesty I would say that this is way out of your skill set .

Walking robots require some very precise engineering and machining not to mention a lot of computer power or even better a lot of computers.


However I don't mean to discourage you but some recommended reading would be ,Physics (university level ) Mathematics (differential calculus ) engineering mechanics and engineering computing , there are a lot of books around on mechatronics, you should also study up a bit on human anatomy and learn the mechanics of walking

This is just a start as you will also need a metal lathe and mill and some fabrication skills to go with them .

I only know basic C++ my self so I am not to sure if that will have the scope for a walking robot

The photos you show look like some ones final year university project in mechatronics .

Personally I would recommend you start with an easier project than a bipedal and get your skills up a bit 

Hi, thank you for responding.  

I've fixed the Youtube video link as well.
Yeah, the robot is very impressive.

Physics, Mathematics, Enginering Mechanic university level is not a problem with me, I'm taking Engineering up to semester 3.

But still not up to the engineering computing and we don't learn Mechatronics. (Civil Engineering)

I'll guess I'll do extra reading on that.  

Btw, can you identify what is the "chip" in the photos I posted?
It even have an SD card slot.
Perhaps to store the program I guess?

What I need help is for the parts.

I know it's quite reckless to start right away with bipedal walking robot, but this is not a short term projects.
It's a hobby and I won't mind even it took me up to 1 year or even more to build it.
It's just for self satisfaction.  

I'm inspired and motivated watching the videos in this Videos section of this forum.

[wgn_white]

The most basic problem with humanoid robots is balancing... ( and ofcourse the fact that they cost too much $$ than some other types of robots)...
so basically , i would say that if you dont have any experience in robotics, dont go for a humanoid just yet (especially if you are doing t all by yourself)....start with a much simpler and basic robot with a differential drive etc....then when you are used to handling servos, and a little complex programming, then you can go for the complex robots..
but please be patient , as learning things takes time...
BEST of LUCK..!!

I'm used to build scratch an RC.

So I can say I'm quite used to servos, etc.
But you know RC is *FAR* from having seems to be intelligent or have a reaction such as robot.

I have the idea for balancing.
Which is gyro.
Helicopter RC using electronic gyro to balance the body in the sky.
As in, when the body tilt on some side, the gyro will detect the changes and "kick" some current to the servos and balance the helicopter.
I am planning to use this theory on balancing, but it's just a theory though.

I will try not to rush it. 

[Soeren]

Hi,

I found the creator website, http://dancingdoll-rz.com/main.htm

It's in Japanese and I can't read those.

Translated (sort of).

[photomark]

Hey Wiguan don't let me or any one else here talk you out of trying this project , you would learn a lot from just thinking about it .


Any good  project starts with a good design and drawing and even more importantly a good design strategy , don't try and design the whole thing at once but instead treat each part as a separate project , for example work on a good knee joint and once you have that worked out and functioning then move to the next part etc etc   
Do you have a good CAD program ? if you don't you can get TurboCad 12.5 deluxe for free and it is very good or you could use google sketch up .

I cant find the link for turbocad 12 but if you do a search you will find it ok ....


That photo of the chip looks to me to be a SD memory card adapter  .

Just for reference on balancing and gyros , the human body has a total of four gyros , two in each ear , a bipedal robot would probably need the same and when we walk we have the equivalent of several thousand computer cores working together to keep us balanced . this is just to give you an idea of the complexity of a walking robot  .

Or if you wish you could buy a walking robot , there is a commercial robot called "NOW" but it cost around the $20,000 mark . you could find it on youtube and it has had a mention on this forum as well .   

 

[wgn_white]

Hi,

I found the creator website, http://dancingdoll-rz.com/main.htm

It's in Japanese and I can't read those.

Translated (sort of).

Haha 
Thanks for that.

Hey Wiguan don't let me or any one else here talk you out of trying this project , you would learn a lot from just thinking about it .


Any good  project starts with a good design and drawing and even more importantly a good design strategy , don't try and design the whole thing at once but instead treat each part as a separate project , for example work on a good knee joint and once you have that worked out and functioning then move to the next part etc etc   
Do you have a good CAD program ? if you don't you can get TurboCad 12.5 deluxe for free and it is very good or you could use google sketch up .

I cant find the link for turbocad 12 but if you do a search you will find it ok ....


That photo of the chip looks to me to be a SD memory card adapter  .

Just for reference on balancing and gyros , the human body has a total of four gyros , two in each ear , a bipedal robot would probably need the same and when we walk we have the equivalent of several thousand computer cores working together to keep us balanced . this is just to give you an idea of the complexity of a walking robot  .

Or if you wish you could buy a walking robot , there is a commercial robot called "NOW" but it cost around the $20,000 mark . you could find it on youtube and it has had a mention on this forum as well .   

 

Hi, photomark.

Thank you for your advise.
I think your advise to work on the project part by part is a very good idea.
So I guess I'll start building the feet first.
I do have Autodesk 10.

Building the robot body is not my main concern, the electronics part is my main concern.

From reading, I've found out about Arduino and sanguino and also some are making their own microprocessors.
This is my great weaknesses.

From the Google translator link that Soeren gave, it seems it's just an Mp3 player and nothing to do with the robot controller, or am I wrong?

NOW is too much for me, that will be an industrial level.

I just wanted to build a 60 cm height robot.

[Webbot]

You could look at Gait Designer which is part of WebbotLib http://webbot.org.uk
It allows you to create a gait (animation) for your servos and is interactive at design time - ie move some sliders on your PC and the servos on your robot move.
You can also play animations.
Once done you can upload it to the microcontroller as a standalone.
You can also vary the elements of a gait at runtime - ie from your balance sensor you could adjust the ankle joints.

Having said all of that - humanoids are really complex - and standard gaits MAY not be the answer as they make assumptions about the environment. In essence human movement is about leaning in the direction you want to go - and then co-ordinating everything else to stop you falling over. A bit like how flying is the art of throwing yourself at the floor but missing!

[wgn_white]

You could look at Gait Designer which is part of WebbotLib http://webbot.org.uk
It allows you to create a gait (animation) for your servos and is interactive at design time - ie move some sliders on your PC and the servos on your robot move.
You can also play animations.
Once done you can upload it to the microcontroller as a standalone.
You can also vary the elements of a gait at runtime - ie from your balance sensor you could adjust the ankle joints.

Having said all of that - humanoids are really complex - and standard gaits MAY not be the answer as they make assumptions about the environment. In essence human movement is about leaning in the direction you want to go - and then co-ordinating everything else to stop you falling over. A bit like how flying is the art of throwing yourself at the floor but missing!

Hi, Webbot.

Thanks for the link.
The Gait Designer seem to do what I plan to do, just like the video above "programming the dance" into the robot.

I have to agree with you that gait's may not be sufficient tool to make a humanoid robot.

I've put my eyes on Kondo robot kit series.

Perhaps I can use these kit electronic parts and modify the chasis, body and frame to become 60cm tall robot.

What do you think about the robot kit?