Legacy Forum: Preserving Nearly 20 Years of Community History - A Time Capsule of Discussions, Memories, and Shared Experiences.

Upgrayd's AX Darwin Project

Dedicated to the new Dynamic Anthropomorphic Robot with Intelligence - Open Platform By Robotis and related projects
10 postsPage 1 of 1
10 postsPage 1 of 1

Upgrayd's AX Darwin Project

Post by Upgrayd » Thu Sep 01, 2011 12:57 am

Post by Upgrayd
Thu Sep 01, 2011 12:57 am

Hey everyone,

I wanted to share the progress I have been making on my AX12 Biped project based on the DARwIn-OP project.

The mechanics of my biped are all custom design and fabricated by me to mimic the joint configuration for the DARwIn-OP.

The main controller for this robot will be a Gumstix Overo COM booted from a micro sd card with and an image of Linux version 2.6.36 and the GNOME desktop. This image also includes native development tools for c, c++ and python.

The software controlling this robot is all custom but heavily based off of the open source code released from the DARwIn-OP sourceforge page.

Image
Overall Picture. From foot the shoulder the robot stands a hair over 1foot tall.

Image
Close up of the feet.

Here are a two videos of the robots fist steps:

http://www.youtube.com/watch?v=tFE8Lseg6IY
Forward walking.

http://www.youtube.com/watch?v=D8nuekcSKko
Sidestepping and turning in place.

----
http://upgraydlabs.com/
http://upgrayd.blogspot.com/
Hey everyone,

I wanted to share the progress I have been making on my AX12 Biped project based on the DARwIn-OP project.

The mechanics of my biped are all custom design and fabricated by me to mimic the joint configuration for the DARwIn-OP.

The main controller for this robot will be a Gumstix Overo COM booted from a micro sd card with and an image of Linux version 2.6.36 and the GNOME desktop. This image also includes native development tools for c, c++ and python.

The software controlling this robot is all custom but heavily based off of the open source code released from the DARwIn-OP sourceforge page.

Image
Overall Picture. From foot the shoulder the robot stands a hair over 1foot tall.

Image
Close up of the feet.

Here are a two videos of the robots fist steps:

http://www.youtube.com/watch?v=tFE8Lseg6IY
Forward walking.

http://www.youtube.com/watch?v=D8nuekcSKko
Sidestepping and turning in place.

----
http://upgraydlabs.com/
http://upgrayd.blogspot.com/
Upgrayd
Newbie
Newbie
Posts: 5
Joined: Wed Feb 17, 2010 3:17 pm

Post by i-Bot » Thu Sep 01, 2011 10:27 am

Post by i-Bot
Thu Sep 01, 2011 10:27 am

That's looking really good.

Are the side steps and turns based on IK and a gait ?
Did you include these into the walking module, or add a new module.

Are the 5 AX12 in the torso mounted direct onto the backplate ?
That's looking really good.

Are the side steps and turns based on IK and a gait ?
Did you include these into the walking module, or add a new module.

Are the 5 AX12 in the torso mounted direct onto the backplate ?
i-Bot
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1142
Joined: Wed May 17, 2006 1:00 am

Post by Upgrayd » Thu Sep 01, 2011 12:52 pm

Post by Upgrayd
Thu Sep 01, 2011 12:52 pm

i-Bot wrote:That's looking really good.

Are the side steps and turns based on IK and a gait ?
Did you include these into the walking module, or add a new module.

Are the 5 AX12 in the torso mounted direct onto the backplate ?


The sidestepping and turning are both part of the walking module all using the same gait and kinematics engine.

The torso servos are all mounted directly to the backplate. The backplate in the images and video is temporary until I can get the material make the aluminum equivalent.
i-Bot wrote:That's looking really good.

Are the side steps and turns based on IK and a gait ?
Did you include these into the walking module, or add a new module.

Are the 5 AX12 in the torso mounted direct onto the backplate ?


The sidestepping and turning are both part of the walking module all using the same gait and kinematics engine.

The torso servos are all mounted directly to the backplate. The backplate in the images and video is temporary until I can get the material make the aluminum equivalent.
Upgrayd
Newbie
Newbie
Posts: 5
Joined: Wed Feb 17, 2010 3:17 pm

Post by nunogato » Thu Sep 01, 2011 10:18 pm

Post by nunogato
Thu Sep 01, 2011 10:18 pm

Hi Upgrayd,

First of all nice post and project :)

You mention that you have custom design and fabricated some parts to mimic the joint configuration of the DARwIn-OP but I see a big difference.

The knee servo on the DARwIn is connected with the upper leg, check this photos
Image
100_2038 by RoboSavvy, on Flickr
Image
100_2033 by RoboSavvy, on Flickr
Image
100_2041 by RoboSavvy, on Flickr

If you need any extra info about the DARwIn let us know
Hi Upgrayd,

First of all nice post and project :)

You mention that you have custom design and fabricated some parts to mimic the joint configuration of the DARwIn-OP but I see a big difference.

The knee servo on the DARwIn is connected with the upper leg, check this photos
Image
100_2038 by RoboSavvy, on Flickr
Image
100_2033 by RoboSavvy, on Flickr
Image
100_2041 by RoboSavvy, on Flickr

If you need any extra info about the DARwIn let us know
nunogato
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 86
Joined: Mon Mar 15, 2010 3:48 pm

Post by Upgrayd » Thu Sep 01, 2011 11:20 pm

Post by Upgrayd
Thu Sep 01, 2011 11:20 pm

Yup, The knee position is different.

I had to change the knee mounting in order to keep the robot as short as possible.

The end result is the same joint to joint proportions as the darwin.
Yup, The knee position is different.

I had to change the knee mounting in order to keep the robot as short as possible.

The end result is the same joint to joint proportions as the darwin.
Upgrayd
Newbie
Newbie
Posts: 5
Joined: Wed Feb 17, 2010 3:17 pm

Post by i-Bot » Fri Sep 02, 2011 1:30 pm

Post by i-Bot
Fri Sep 02, 2011 1:30 pm

As Upgrayd says the difference is minor. The overall joint layout is the same since it is only the axis which is important not the servo position.

If you conform to the general Darwin OP joint layout described here:
http://robosavvy.com/site/index.php?opt ... 1&Itemid=2
then changes to the software are parametric changes for the different dimensions and some setting of servo directions and zeros.

If you diverge from the Darwin-OP layout, then more work is required to modify the IK code.

The other thing is joint constraints, though that is more important form complex moves like getting up, rather than walking.
As Upgrayd says the difference is minor. The overall joint layout is the same since it is only the axis which is important not the servo position.

If you conform to the general Darwin OP joint layout described here:
http://robosavvy.com/site/index.php?opt ... 1&Itemid=2
then changes to the software are parametric changes for the different dimensions and some setting of servo directions and zeros.

If you diverge from the Darwin-OP layout, then more work is required to modify the IK code.

The other thing is joint constraints, though that is more important form complex moves like getting up, rather than walking.
i-Bot
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1142
Joined: Wed May 17, 2006 1:00 am

Post by Tyberius » Thu Sep 22, 2011 5:45 pm

Post by Tyberius
Thu Sep 22, 2011 5:45 pm

i-Bot wrote:The other thing is joint constraints, though that is more important form complex moves like getting up, rather than walking.


Just to confirm, this is actually set via CCW & CW angle limits in the registers of the dynamixels, correct? I noticed that the dynamixel firmware upgrade script sets this for each ID.
i-Bot wrote:The other thing is joint constraints, though that is more important form complex moves like getting up, rather than walking.


Just to confirm, this is actually set via CCW & CW angle limits in the registers of the dynamixels, correct? I noticed that the dynamixel firmware upgrade script sets this for each ID.
Tyberius
Savvy Roboteer
Savvy Roboteer
Posts: 108
Joined: Tue Nov 27, 2007 6:57 pm

Post by i-Bot » Thu Sep 22, 2011 7:40 pm

Post by i-Bot
Thu Sep 22, 2011 7:40 pm

As you diverge from the standard DARwIn OP build then the physical joint angle constraints will change. the first stage is to measure the actual joint constraints of your configuration. Some allowance needs to be made for protection and consideration for how the legs interact. The objective here is to set CW and CCW limits to give best servo protection.

While the MX firmware installer has hardcoded limits which are set with firmware updates, this may not work for AX based robots. The dxl_monitor also sets the CW and CCW limits on RESET. Again these are hardcoded and must be changed to meet your configuration above.

There may also be value to adding some debug code to the action or walking modules to flag if limits are ever exceeded in custom configurations.
As you diverge from the standard DARwIn OP build then the physical joint angle constraints will change. the first stage is to measure the actual joint constraints of your configuration. Some allowance needs to be made for protection and consideration for how the legs interact. The objective here is to set CW and CCW limits to give best servo protection.

While the MX firmware installer has hardcoded limits which are set with firmware updates, this may not work for AX based robots. The dxl_monitor also sets the CW and CCW limits on RESET. Again these are hardcoded and must be changed to meet your configuration above.

There may also be value to adding some debug code to the action or walking modules to flag if limits are ever exceeded in custom configurations.
i-Bot
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1142
Joined: Wed May 17, 2006 1:00 am

Post by MOHIT JINDAL » Tue Nov 01, 2011 3:23 pm

Post by MOHIT JINDAL
Tue Nov 01, 2011 3:23 pm

Hi Upgrayd,
I saw your blog. Can you tell me from where you got aluminium frames ? :roll:
Hi Upgrayd,
I saw your blog. Can you tell me from where you got aluminium frames ? :roll:
MOHIT JINDAL
Savvy Roboteer
Savvy Roboteer
Posts: 178
Joined: Wed Nov 10, 2010 7:43 am

Post by Upgrayd » Wed Nov 02, 2011 2:43 am

Post by Upgrayd
Wed Nov 02, 2011 2:43 am

MOHIT JINDAL wrote:Hi Upgrayd,
I saw your blog. Can you tell me from where you got aluminium frames ? :roll:


I designed the brackets in a 2D cad program, had them cut by a waterjet service and bent them on a metal brake.
MOHIT JINDAL wrote:Hi Upgrayd,
I saw your blog. Can you tell me from where you got aluminium frames ? :roll:


I designed the brackets in a 2D cad program, had them cut by a waterjet service and bent them on a metal brake.
Upgrayd
Newbie
Newbie
Posts: 5
Joined: Wed Feb 17, 2010 3:17 pm


10 postsPage 1 of 1
10 postsPage 1 of 1