Older blog entries for Kiryu (starting at number 2)

Ah, so I have learned the lesson of tolerances! I had to remove approximately .1mm of material from one of the shoulder pieces. I guess the .5mm tolerance that I had before was not small enough. I also have learned that I will need smaller bolts to attach the servo arm to the aluminum; the heads on the bolts were too big and ran into the shaft portion of the servo connector. This is okay, it just means another trip to the local hobby store, since there seems to be no other store around that carries a small enough bolt. The regular JB weld is working great. I let it sit overnight and it hardened as expected. I will probably only use the quick version in places that have large surface contact areas (as in the upper arm). Got to love this epoxy! I am impressed with the repeatability in which the mill is able to make parts. The parts that I machine are almost identical (by almost, I mean .01mm difference). That is one good machine! So my dead parts pile is getting larger. As a design is passed up for a new one or a part is machined incorrectly they get thrown there. This should make an interesting picture at the end of the skeleton building phase. I am off to machine the other shoulder now. There will most likely not be an update over the weekend, I will be off to work (like everyday it seems) and my spare time will be taken up by DJing on saturday and flying model airplanes with my stepdad (you have to do something to keep in touch after you move away) on sunday. Hope your projects are going well and if you hvae read this far I appreciate the interest!

7 Sep 2001 (updated 7 Sep 2001 at 03:49 UTC) »

I am building the shoulder interconnect today. Being a generational leap over the Gobo project, I actually have tollerances to workout :-) Luckly, the Sherline mill is here to help me. In this particular part, .2mm is the max deviation for some of the pieces. This resulted in some discarded pieces and an extra hour or so of work. I am still using JB Weld as the epoxy of choice, but I am attempting to use the quick version (5 min set opposed to 3 hour set). Although it worked on the porous plastic of the servos to bind together the upper arm, the well machined aluminum wasn't sticking very well. I scored it with a dremel in hopes to get it to set better. If that doesn't work, I will ge back to the 4 hour JB weld. I highly recommend JB weld to anyone using aluminum as a building material (less the building of a battlebot). It saves on space (fewer bolts) and money (fewer bolts :-) So that is the state of the humanoid robot. By the way, I received quite a few emails in regards to Gobo. I wish I had plans to give everyone, but I don't have them in any conventional format. Quite honestly, Gobo was the result of a LOT of labor on each individual part. Each part was cut to specifications with a dremel too (pre-Sherline Mill era), which was very laborous at times. Using the mill, I could probably get it built in 1/3 of the time. Depending on how may more requests I get, I may throw together a article on Gobo. If you have no idea what I am talking about, you can see pictures of Gobo at the TRCY website www.trcy.org under the "My Quadraped" section. Until next time..... oh, and the JB Quick didn't work as well as I hoped so it is on to the regular (which was used on Gobo)

Things are getting underway with my new project, a humaniod torso. To give you a little history, I have designed a robot "dog" (actual mesurements were taken from a gorrilla) named "Gobo". Once I was finished with the frame, every toy company in the world came out with a robot dog. Just before I wrote a 3d motion file exporter for Milkshape 3D, I stopped the project. Pictures can be viewed in the TRCY robot club under the My Quadraped section. The new project is a humanoid torso at 1/10th scale of my body measurements. I am using micro Hitec servos for the actuators. The servo controller will be a custom design based on the Atmel 8535 (or mega 103), which will be an extension of the 2 that were used on Gobo. Controlling high level function will be a Rabbit 2000 core microcontroller, which will filter motion data down to the servo controller. The Rabbit will accept vision information from a Atmel Mega 103 and GB Camera vision system. Voice is also planned as an extension of this project. Currently, I have managed to alter some of the servos for my purpose and started design on the torso frame. Machining is being done with a Sherline mill. The first phase will be completion of the frame and installation of the servo controller. The second phase will be to get the servo controller and Rabbit 2000 playing nicely together :-) and for basic motion data to be transmitted. The Third phase will be the completion of a motion file exporter for Milkshape 3D and motion data handler for the Rabbit 2000. The fourth phase is the integration of the vision system and the ability to recognize simple objects (circles that are really bright etc). This will be an ongoing process as the needs for the vision system chages. Voice synthesis is planned to allow for the robot to say canned (stored) voice streams. Once this is complete, the true use for the system will begin :-) If you are interested in an aspect of this project or would like to make a suggestion you can email me at kiryu69@yahoo.com.

Nathan

X
Share this page