PDA

View Full Version : Arbotix based Quad (unnamed)



DC Robotic
10-18-2009, 07:34 PM
Here are initial pictures of my first walker (Quad).

It uses an Arbotix-xbee system and 14 AX-12s. 3000ah 3s lipo for power. Bellcrank driven linear bearing "feet". These allow the robot to stand and support considerable weight unpowered. The chassis is mocked up from birch ply right now. Should be able to carry lots 'o payload with the leg design. Trendnet working but not mounted and No guns yet. The Arbotix is up and running and it has taken a few steps. I'd name it LURCH but I have plans to develop a smooth gait eventually. Essentially the legs are 2dof + (up-down) so the eventual kinematics should be fairly simple to implement. The next steps will be experimenting with the Arbotix and planning a control panel.


http://forums.trossenrobotics.com/gallery/files/2/7/1/4/img_0002.jpg

DC

lnxfergy
10-18-2009, 10:07 PM
Need.. close up... of.. leg. That's really cool looking

-Fergs

jes1510
10-18-2009, 10:43 PM
I'm inclined to say the bell crank idea is brilliant. I love it!

sthmck
10-18-2009, 11:18 PM
Looks great, nice idea with the legs.

Zenta
10-19-2009, 04:08 AM
Hi,

Very creative solution for the legs! You should be able to carry some bricks on that quad.
The design are very equal to this one:


http://www.youtube.com/watch?v=MtU7FlR1X14

Agree with Fergs, some more closeups would be appreciated.
I hope you are willing to share your code ;).

(I'm also waiting for the ArbotiX since my comprehensive kit has been to a very little use..:()

-Zenta

sam
10-19-2009, 09:33 AM
[COLOR=black][FONT=Verdana]3000ah 3s lipo for power.


You about that? ;)

The design is really AWESOME! You'll be able to have lots of payload! This is inspiring...

Sam

Upgrayd
10-19-2009, 10:12 AM
Very solid design you have there. Will be fun to see this one progress.

WTB closeup pictures of legs!

DC Robotic
10-19-2009, 11:49 AM
By request. Here are some close-up's of the leg design.

@ Fergs. Here are a couple of close-ups on the legs. Had the micro linear rails from a ebay purchase for cnc router parts. The Actuator rods are RC helicopter parts. The aluminum brackets I fabricated and the servo horns are cut from white UHMW. They are tall so the rods remain in-plane with the rails.


Thanks for the compliments Jes, sthmck.


Zenta, OH YEA, Thatís the way I envision it walking. Wish I had a tenth of your IK skills! Right now Iím planning an approach to the software. Iím thinking Iíll write a control program in C# which reads joystick data-then writes poses/header files out to Fergs pypose on the Arbotix. Iíll probably stick to C# as I already have a pose-capture program started that I used with USB2dynamixel that was working fairly well. And I also have a joystick reading program. It looks like it would be very easy to adapt the output for the Arbotix.


Sam, oops, 3000 ďmilliampĒ hrs. wish it was 3000ahÖIíd be building an electric vehicle to crush a Tesla with!


Showing down, mid, and up positions. The black lines on the right rail show the limits of travel. I could increase the travel a bit, but at tthe time I did not have the patience to drill the hardened rails.
http://forums.trossenrobotics.com/gallery/files/2/7/1/4/legscloseupsm.jpg
Super close up of the mechanism.
http://forums.trossenrobotics.com/gallery/files/2/7/1/4/superclosesm.jpg

The naked underbelly of the beast. I wanted the lengths of the "thigh" and "femur" to be similar for the greatest range of motion. I want to remake the "knee" bracket to allow a more acute angle.
http://forums.trossenrobotics.com/gallery/files/2/7/1/4/underbellysm.jpg

rebel
10-19-2009, 01:05 PM
Nice design!

Zenta
10-19-2009, 01:55 PM
Nice closeups! I loved your use of the micro linear rails, excellent!

Have you considered making the "feet" larger so that they have a wider contact area with the floor? I believe large feet would make it much easier to walk in a stable gait while lifting two legs at a time. Since this leg design doesn't allow the robot to do properly roll or pitch rotations (only yaw rotation) large feet wouldn't be a trouble. The kinematics you need for the horisontal hip and "femur" are the same I used for the femur and tibia joints in the PEP sheet for controling Phoenix or the ik calcs Xan used in his BAP code for Phoenix. So it should be a relative easy task to solve the 2DOF kinematics. Since you only need body rotation along one axis (yaw roation) also these calculations are easier to do, no need for a rotation matrix.

Keep up the good work!

-Zenta

DC Robotic
10-19-2009, 02:52 PM
Zenta, Thans for the encouragement. Feet are on the drawing board. You are correct that they will help substantially. I'm thinking they will be like shallow arches (points down) that wil help resist tipping by being nominally oriented perpendicular to an imaginary line between opposite feet. I'm pretty green with IK but i do know that 2dof can be simply solved with trig. I'll be researching the solutions you mentioned.