First Attempt to print out DARwIn-OP brackets!

[Gort]

We printed this out on a Ultimaker

http://blog.ultimaker.com/



http://mike-ibioloid.blogspot.com/2011/ ... n-3-d.html

[limor]

Having played around with the Makerbot Thing-O-Matic for a few months and seen some great looking results, this bracket printed by ultimaker looks pretty ..ehm.. bad.


I'd like to bring up something we've been thinking about which is is to adapt the DaRWin brackets to AX12 servos.

It wont have fancy magnetic encoders and ability to lift an elephant but it should be pretty cool to have a low cost version of the super humanoid and leverage the efforts that are currently being put into software for DaRWin.

[i-Bot]

You don't need to adapt DARwIn Brackets to AX12 servos, the standard AX12 brackets are fine. The only differences from a premium is simple rotation of the elbow, and integration of the head.

For the head, you can:
1) adapt the body for the head pan servo
2) fit the head pan and head tilt servo in the head
3) stick with 18 DOF without head

Of course you may want to make cosmetic changes too.

After that all other changes can be made in software.

[limor]

Hi i-Bot,

We'll be receiving the DaWIn in a few weeks. hopefully before then we'll be able to print out some adapted parts to make a bioloid Darwin chimera.

Anyone else working on this ?

[Gort]

Yes

[billyzelsnack]

The Darwin CAD models really are amazing. All the parts are there. The body shell is quite large and will be difficult to get a quality print. Especially difficult because its curves require a LOT of support. I'm going to use it as a test benchmark for the printer I'm building. In the mean time I'll probably do a print at half scale on the cupcake tonight and see how it comes out.

[i-Bot]

Hi Limor,

Is your goal to achieve a look alike, or to have the same functions ?

If functions which of these do you consider is essential ?

Pan Tilt Video camera ?
Linux processor ?
Gyro ?
Accelerometer ?
Sound ?
Head and Eye LED ?
Force sensing resistors ?

The major part of the design is the torso and head. The above functions have most impact on that.

It is better to keep as close as possible to the Bioloid premium on joint layout, so you don''t have to make new motion files for the moves other than walking. Walking is with the DARwIn walking module, other moves are created in Roboplus.

[limor]

Pan Tilt Video camera ? - yes
Linux processor ? - yes + wifi
Gyro ? - yes - same as Darwin
Accelerometer ? - yes same as Darwin
Sound ? - nice to have
Head and Eye LED ? - nice to have
Force sensing resistors ? - where ? the CM510 can collect sensor input and the linux board can get additional sensors via an I/O board such as arduino attached to one of its USB ports.


The torso and head will need some redesign to adapt to a non fitPC2 computer and to CM510. for ball tracking and simple openCV and inverse dynamics gait generators + wifi, the DMP boards should be more than enough.

The weights and inertias will have to be recalculated in order for the dynamic walk generators to work. original gaits from premium may work.. maybe.

printing the smooth outer shells is a challenging with Markerbot or similar 3D printers. maybe printing them with Makerbot and then using them as molds for fibreglass to create the actual shells should get better results.

btw: the linux will connect to the CM510 via the zigbee port and ftdi because the RS232 is limited to 225kbps. if i recall from CM5 dissection times, the zigbee port on the CM510 is converted to 3.3V but should not have any speed constraints. right?

[i-Bot]

If you retain the CM-510, this probably prevents the head pan servo being in the torso, so you need a head design with both pan and tilt. plus camera.

Accelerometer and gyro are best direct on DMP if you use CM510.

The robotis code does not use dynamic walk, so only kinematics need analysis, not dynamics.

It is not the gaits of the premium which are of value, but the get up routines and similar.

I don't have CM-510, but you may look at using with LN-101.

[billyzelsnack]

Well that turned out about as expected! haha.

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

At 50% this thing is almost a worst case scenario for a FDM printer with sparse supports. Everything is a single thread thick with lots of overhangs and supports. My cupcake bed is too small to print the body out at 100%, but maybe I'll give a try to chopping it into 4 and printing at full scale.

[Gort]

We think the Ultimaker can easily do it. My friend who owns the printer is fine tuning it now. The two brackets we printed where his first major prints. It is just a matter of time before we are printing out body shell parts!

[Gort]

My friend Luis just print out this bracket last night with his Thing-O-Matic! He said it was as strong as the brackets on my robot!



http://www.thingiverse.com/derivative:10965

[Luis Rodriguez]

Yes, be kind to my Ultimaker friend, he is still tuning Skeinforge, we are using this project as an excuse to fine tune. I've done pretty well, I still am putting down a little too much plastic but this part is definetly Usable and will compete the entire Darwin in this awesome UV Red. I'm not sure I have the scale right just yet but that is easily fixable. Anyways, the race is on!

This Flickr set shows how clean it was after the print was finished, if I do say so myself.

http://www.flickr.com/photos/luiserodri ... 007269603/

I'll keep working on it tonight and tomorrow evening. Brackets dont' seem to be the problem, the head and body will be fun!

[billyzelsnack]

Looks great.

I was inspired to print one out myself. Once again I am seeing the same problem with printing brackets. The side wall strength is pretty poor because of layer delamination. I really want to figure out why. Yeah my print has other various issues, but right now I am more concerned about part strength. Any ideas on why your part is so much stronger than mine?

http://www.youtube.com/watch?v=e0YZ_9ERO7s

BTW. How wide did your bracket sides end up printing out? The model is 2mm and mine ends up at 2.32mm.

[Luis Rodriguez]

Looks great. What layer height?

.3 and I'm using the .4 nozzle with the cartridge heater. (Mk6+) That naming convention is starting to get silly.

I use 5D version of RepG 24 (Skeinforge 40) which you just put in the filament diameter (Dimension) and that is about it. You can adjust speed and retraction/reversal but I no longer monkey with W/T values.

I am going to try Rep G 25 though. I had the best out of the box experience with Rep G 24's default profiles so I am curious.

5D Rep G 24 - https://github.com/giseburt/ReplicatorG

Luis