Building humanoid similar to Asimo and Qrio

RoboSavvy distributes and manufactures robots. This forum is dedicated to robots and other bits designed or manufactured by RoboSavvy and robot software developed by RoboSavvy.
127 postsPage 1 of 91, 2, 3, 4, 5 ... 9
127 postsPage 1 of 91, 2, 3, 4, 5 ... 9

Building humanoid similar to Asimo and Qrio

Post by limor » Wed Jul 25, 2012 10:38 pm

Post by limor
Wed Jul 25, 2012 10:38 pm

I've been playing with humanoid robots now for several years and finally I realized that it is time for me to design one.

I want it to look like Asimo and Qrio but to use standard hobby robot servos to make it affordable to hobbyists and AI researchers.

So here's my inspiration:

Image

Image
I've been playing with humanoid robots now for several years and finally I realized that it is time for me to design one.

I want it to look like Asimo and Qrio but to use standard hobby robot servos to make it affordable to hobbyists and AI researchers.

So here's my inspiration:

Image

Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by limor » Wed Jul 25, 2012 10:52 pm

Post by limor
Wed Jul 25, 2012 10:52 pm

To start the project I'm focusing on the hips.
Asimo and Qrio set a standard for humanoid robot aesthetics.

if we look at the legs from within the torso then we see that the Roll DOF actuator is hidden in the torso. The Yaw and Pitch are in well occluded Universal Joint format. The "skin" that occludes the Universal Joint also limits the range of motion. This is not an issue because the robots don't need to have a range of motion wider than that of an average human.


If we look at other humanoid robot hip designs, some of the exposed servos are not pretty in comparison to the Asimo/Qrio. Some robots like Darwin-OP have partial skinning and others have full skinning. NAO's skin is ok except for the hip where you can see through the Universal Joint.

Image
To start the project I'm focusing on the hips.
Asimo and Qrio set a standard for humanoid robot aesthetics.

if we look at the legs from within the torso then we see that the Roll DOF actuator is hidden in the torso. The Yaw and Pitch are in well occluded Universal Joint format. The "skin" that occludes the Universal Joint also limits the range of motion. This is not an issue because the robots don't need to have a range of motion wider than that of an average human.


If we look at other humanoid robot hip designs, some of the exposed servos are not pretty in comparison to the Asimo/Qrio. Some robots like Darwin-OP have partial skinning and others have full skinning. NAO's skin is ok except for the hip where you can see through the Universal Joint.

Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by limor » Wed Jul 25, 2012 11:31 pm

Post by limor
Wed Jul 25, 2012 11:31 pm

My first take on having a Universal Joing hip involved using the standard Bioloid brackets + custom 3D printed cross + pulleys and belts

Image
My first take on having a Universal Joing hip involved using the standard Bioloid brackets + custom 3D printed cross + pulleys and belts

Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by limor » Thu Jul 26, 2012 12:07 am

Post by limor
Thu Jul 26, 2012 12:07 am

Here's what it would kind of look like with two legs and 3 servos.

Everything above the Universal Joint will be hidden under the torso skin.

The Yaw rotation is direct driven by the top servo which will be linked to an aluminum chest skeleton.

The Pitch and Roll rotation are driven by servo linked to pulleys and belts

Just above the Universal Joint there will be a ball bearing that is linked to the same aluminum chest skeleton.

Image
Here's what it would kind of look like with two legs and 3 servos.

Everything above the Universal Joint will be hidden under the torso skin.

The Yaw rotation is direct driven by the top servo which will be linked to an aluminum chest skeleton.

The Pitch and Roll rotation are driven by servo linked to pulleys and belts

Just above the Universal Joint there will be a ball bearing that is linked to the same aluminum chest skeleton.

Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by limor » Thu Jul 26, 2012 12:48 am

Post by limor
Thu Jul 26, 2012 12:48 am

Then I decided that these belts were too thick and we came up with an alternative. Since we don't need any elasticity in the transmission of the servo shaft to the Universal Joint, we decided to use fishing rods. We'll see how that fairs out.

Image
Then I decided that these belts were too thick and we came up with an alternative. Since we don't need any elasticity in the transmission of the servo shaft to the Universal Joint, we decided to use fishing rods. We'll see how that fairs out.

Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by MOHIT JINDAL » Thu Jul 26, 2012 3:26 am

Post by MOHIT JINDAL
Thu Jul 26, 2012 3:26 am

:D :lol: 8) Very Good. That is my Dream from 2004. If you read all my post threads, you will notice I am only making a Humanoid robot like ASIMO. Look here:-
http://www.flickr.com/photos/mohitjindal2712
And watch the video http://www.youtube.com/watch?v=lX0uAKmYLgA
The biggest problem I have is I am making it alone. No Team and No funds from university. No machines to cut 100% accurate.
Second big problem is High torque Servos are very costly; so I started the thread "Servos from DC motor". Which not helped me actually.
Third problem is I am making it practically. No design on CAD or any simple software.
ASIMO uses Harmonic Drives that are far more costly. Only a company can afford it.
I hope we will get success one day before 21st Dec 2012 :wink:
:D :lol: 8) Very Good. That is my Dream from 2004. If you read all my post threads, you will notice I am only making a Humanoid robot like ASIMO. Look here:-
http://www.flickr.com/photos/mohitjindal2712
And watch the video http://www.youtube.com/watch?v=lX0uAKmYLgA
The biggest problem I have is I am making it alone. No Team and No funds from university. No machines to cut 100% accurate.
Second big problem is High torque Servos are very costly; so I started the thread "Servos from DC motor". Which not helped me actually.
Third problem is I am making it practically. No design on CAD or any simple software.
ASIMO uses Harmonic Drives that are far more costly. Only a company can afford it.
I hope we will get success one day before 21st Dec 2012 :wink:
MOHIT JINDAL offline
Savvy Roboteer
Savvy Roboteer
Posts: 178
Joined: Wed Nov 10, 2010 7:43 am

Post by limor » Thu Jul 26, 2012 9:22 am

Post by limor
Thu Jul 26, 2012 9:22 am

CAD is essential not only to design the parts but also to understand the mechanical limitations of the systems. If it doesn't work in CAD then you can easily fix the design.

btw: I'm guessing that your knee servo will not be able to sustain walking because as the knee bends, the torque on the knee joint increases at around COS(angle)* [weight of upper leg + torso]*[upper leg length to center of mass]. so if your upper leg length is about 30cm and the knee-up weight is around 1.5kg (after you added battery, computer and arms, head), when the robot lands on one foot when walking and the knee is bent at 60deg, the torque is 1.5*30*COS(60) = 22.5 kg-cm and that's without inertia which can add another few kg force. so even two AX12 may not do the job assuming the legs are that long.

As for fabrication, we have (at RoboSavvy lab) 3D printers, CNC, reflow and soon vacuum forming and maybe plastic molding. You can get access to fabrication machines by joining a local hackerspace. The cost of materials is not so big when you consider the hours you spend on your hobby :)
CAD is essential not only to design the parts but also to understand the mechanical limitations of the systems. If it doesn't work in CAD then you can easily fix the design.

btw: I'm guessing that your knee servo will not be able to sustain walking because as the knee bends, the torque on the knee joint increases at around COS(angle)* [weight of upper leg + torso]*[upper leg length to center of mass]. so if your upper leg length is about 30cm and the knee-up weight is around 1.5kg (after you added battery, computer and arms, head), when the robot lands on one foot when walking and the knee is bent at 60deg, the torque is 1.5*30*COS(60) = 22.5 kg-cm and that's without inertia which can add another few kg force. so even two AX12 may not do the job assuming the legs are that long.

As for fabrication, we have (at RoboSavvy lab) 3D printers, CNC, reflow and soon vacuum forming and maybe plastic molding. You can get access to fabrication machines by joining a local hackerspace. The cost of materials is not so big when you consider the hours you spend on your hobby :)
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by limor » Thu Jul 26, 2012 10:11 am

Post by limor
Thu Jul 26, 2012 10:11 am

Preparing parts for 3D printing using molten plastic deposition (Makerbot) is a challenge on its own.

After several attempts, I finally managed to get the cavities right. Turns out that about 0.3mm additional diameter usually does the job. In this case I printed half a Universal Bracket to see that the 2 AX12 plastic ring washer and bearing fit correctly, in addition to the small hexagonal nut.

Image
Preparing parts for 3D printing using molten plastic deposition (Makerbot) is a challenge on its own.

After several attempts, I finally managed to get the cavities right. Turns out that about 0.3mm additional diameter usually does the job. In this case I printed half a Universal Bracket to see that the 2 AX12 plastic ring washer and bearing fit correctly, in addition to the small hexagonal nut.

Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by MOHIT JINDAL » Thu Jul 26, 2012 10:33 am

Post by MOHIT JINDAL
Thu Jul 26, 2012 10:33 am

Thanks for the Torque calculation for me 8) Can you tell me the torque on the Ankle ? One leg is 60cm. I made same as ASIMO.
Fishing Rods are those to catch fish ? :lol:
I am waiting for the CAD photos with fishing rods.

Here is a Inside ASIMO pic for you:-

Image
Thanks for the Torque calculation for me 8) Can you tell me the torque on the Ankle ? One leg is 60cm. I made same as ASIMO.
Fishing Rods are those to catch fish ? :lol:
I am waiting for the CAD photos with fishing rods.

Here is a Inside ASIMO pic for you:-

Image
MOHIT JINDAL offline
Savvy Roboteer
Savvy Roboteer
Posts: 178
Joined: Wed Nov 10, 2010 7:43 am

Post by nicolas gomez » Thu Jul 26, 2012 3:52 pm

Post by nicolas gomez
Thu Jul 26, 2012 3:52 pm

limor wrote:Then I decided that these belts were too thick and we came up with an alternative. Since we don't need any elasticity in the transmission of the servo shaft to the Universal Joint, we decided to use fishing rods. We'll see how that fairs out.

Image


Hi limor!! here some links may be help

http://projetromeo.com/romeo-documentat ... _mass.html

http://projetromeo.com/romeo-documentat ... matic.html

http://projetromeo.com/romeo-documentat ... joint.html
limor wrote:Then I decided that these belts were too thick and we came up with an alternative. Since we don't need any elasticity in the transmission of the servo shaft to the Universal Joint, we decided to use fishing rods. We'll see how that fairs out.

Image


Hi limor!! here some links may be help

http://projetromeo.com/romeo-documentat ... _mass.html

http://projetromeo.com/romeo-documentat ... matic.html

http://projetromeo.com/romeo-documentat ... joint.html
nicolas gomez offline
Savvy Roboteer
Savvy Roboteer
Posts: 95
Joined: Sat Aug 08, 2009 2:40 pm

Post by nunogato » Thu Jul 26, 2012 4:53 pm

Post by nunogato
Thu Jul 26, 2012 4:53 pm

thanks for the links!
that's one heavy robot.
they seem to have some spring coupling the righ and left side of the hips?

Image
Image
thanks for the links!
that's one heavy robot.
they seem to have some spring coupling the righ and left side of the hips?

Image
Image
nunogato offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 86
Joined: Mon Mar 15, 2010 3:48 pm

Post by mog123 » Sat Jul 28, 2012 2:14 pm

Post by mog123
Sat Jul 28, 2012 2:14 pm

Hey limor,

great looking work!

Check out how hips are made in the RE F60A futaba robot - it might be helpful, it's quite interesting how they did it:

http://imgur.com/a/cFlxx
Hey limor,

great looking work!

Check out how hips are made in the RE F60A futaba robot - it might be helpful, it's quite interesting how they did it:

http://imgur.com/a/cFlxx
mog123 offline
Savvy Roboteer
Savvy Roboteer
Posts: 50
Joined: Sat Jan 30, 2010 2:19 pm

Post by limor » Sat Jul 28, 2012 5:17 pm

Post by limor
Sat Jul 28, 2012 5:17 pm

I've never seen anything like that!
these look like configurable universal joints based on a ball and unusual looking servo

Image
I've never seen anything like that!
these look like configurable universal joints based on a ball and unusual looking servo

Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by limor » Sat Jul 28, 2012 5:19 pm

Post by limor
Sat Jul 28, 2012 5:19 pm

I've made some progress today.
The legs are starting to shape up with my Universal Joint setup.
Here i've done a rough sketch of a polycarb cover similar to Asimo setup.

Image
Image
I've made some progress today.
The legs are starting to shape up with my Universal Joint setup.
Here i've done a rough sketch of a polycarb cover similar to Asimo setup.

Image
Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Post by limor » Mon Jul 30, 2012 2:46 pm

Post by limor
Mon Jul 30, 2012 2:46 pm

The knee has to allow bending to a relaxed position on the shins.
Stress on this bridge should be ok in 3D printed plastic because I designed this knee bridge so that it can be 3D printed so that slices are C shaped and no dramatic mid air angles.
Not too sure how this alignment of servos along the leg will influence walking

Image

Image

Image

Image
The knee has to allow bending to a relaxed position on the shins.
Stress on this bridge should be ok in 3D printed plastic because I designed this knee bridge so that it can be 3D printed so that slices are C shaped and no dramatic mid air angles.
Not too sure how this alignment of servos along the leg will influence walking

Image

Image

Image

Image
limor offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1845
Joined: Mon Oct 11, 2004 1:00 am
Location: London, UK

Next
Next
127 postsPage 1 of 91, 2, 3, 4, 5 ... 9
127 postsPage 1 of 91, 2, 3, 4, 5 ... 9