Designing the body for a Keepon

(I've also posted this to the Seattle and RSSC lists, so my apologies if you get it more than once!)
I'd like to try building a Keepon:
http://beatbots.org/research /
http://www.youtube.com/watch?v=CO3_CVwXCos

http://www.youtube.com/watch?v=ewGk7aKjk7k

http://www.youtube.com/watch?v=lpUz1M0DHZU

(In the last video, I presume it's teleoperated. I don't think it can tell when it's being scolded or applauded. In any case, I'm only planning to do as much as I have software for -- beat detection, face detection, and tracking.)
I'm looking for ideas for the body design, esp. in terms of motion.
It looks like there are 4 motion primitives: - vertical up and down - pan - tilt - side-to-side rocking
Have I missed any?
My mechanical skills and equipment are limited (to say the least :) so ideally, I'd like to use something like the Lynxmotion servo erector set for this -- something I can essentially bolt together. Does anyone have experience using these? Do you think that's a feasible approach (precise enough, compact enough)?
Looking at their parts, I think I can see how everything except bobbing up and down can be accomplished. I'm a little stumped on that one! The two things that comes to mind are 1) a piston, cylinder, crankshaft assembly, which seems way too elaborate, or 2) some sort of cable and tube arrangement. But, I don't see parts for building a cable-based mechanism in the erector set. Did I just miss that?
Thanks for any ideas or suggestions you might have on this! Robin
Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Hey, in case anyone is thinking about this, just want to let you know that Alan Marconett posted an idea on the SRS list that I plan to try as my first attempt: marionette action in which strings pull on the body from below. There's also a blurb in HackedGadgets confirming that something like this is how the Keepon works. Once I get a start on trying to build this, I'll post an update.
Cheers! Robin
On Feb 23, 6:19 pm, snipped-for-privacy@gmail.com wrote:

Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks, Roy!
I have it! I'm pretty sure I've been through all their publications. The HackedGadgets photo and blurb went over the same ground, but had more construction-oriented detail, which was why that one was especially helpful!
Robin

Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Feb 23, 9:19 pm, snipped-for-privacy@gmail.com wrote:

The History Channel video (2nd one from top) has a clip explaining the mechanics: http://beatbots.org/category/videos /
Are you able to open source the software if there were an open source hardware design?
Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Oh my gosh, I can't believe I missed this one!! I actually went through every search hit on youtube until I stopped seeing yellow, did a google image search, and read all the papers from this cmu group, yet failed to scroll down to the second video on beatbots. DOH! It's excellent. Thank you!

I'm not sure what you're asking. Are you making an offer to do an open source hardware design in exchange for open source software?
Thanks! Robin
Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Feb 25, 10:24 pm, snipped-for-privacy@gmail.com wrote:

More like collaboration rather than exchange. It seems like everybody wants one of these things. The mechanical stuff isn't too difficult but people will be very disappointed if they build the thing but don't have the software tools to give it personality. It seems like a perfect project that could benefit from open source collaboration on the hardware and software side. What would people want in a motion editing tool? - loading MP3s, viewing sound/motion waveforms, motion sliders, face tracking, file sharing, etc.
I have a lot of projects going on but I like mechanics the most when it comes to robots. My goal would be to introduce a basic mechanical design that sparks interests and then maybe other folks could take over.
Also, the "Keepon" like creature probably shouldn't be an exact replica. I just emailed the company to see if they would open source the hardware and software design but I'm not expecting a positive (if any) response.
Danh
Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Hi Dan,
I think you're right that the mechanical probably isn't too hard. That's why I picked it. It might even be simple enough for me. (Well, we'll see!!) That plus it's so freakin cute! I'm definitely not planning to just copy. I couldn't anyhow -- theirs is too slick. I'm planning to use just a styrofoam ball for the head and have the foam body itself provide springiness to tension the marionette strings. Other differences: I plan to allow bobbing only in the upright position (more lifelike), use just one camera (at the nose), and put the mic in the base. Plus, I'll try with servos instead of DC motors. Oh, and mine might (eventually) end up looking more like a baby penguin than a baby chicken, or have ears that perk forward, or a comb that lifts....
If CMU (or Beatbots) does an open-source release, you certainly won't need anything from me :))
About you and I starting an open-source collaboration, well we can discuss more detail about that, but my main interest at the moment is finding and building a (fun) mechanical platform I can use to evaluate and to further develop my software.
At some point, I'd like to find a business model that allows me to continue developing software like this and ideally even hire people to participate. I don't know yet what that business model would look like, but I want it to be something I can fall in love with. I don't want to burn myself out on something that doesn't excite me -- and money per se doesn't do that for me. So for now, I'm just pursuing my joy, and seeing where that leads. Meanwhile, while I'm not necessarily opposed to giving away years of effort, I'd like to do that in a way that preserves my options. For example, I'm already inclined to give away the face following software. But I'll probably do so in binary form -- at least initially.
I hope this doesn't come across as unfriendly. I don't mean it that way. It's just timing: I'm not ready to think about the prospect of setting up and managing -- and possibly becoming very frustrated by -- an open source effort. (I went through the latter experience fairly recently, so I'm a little gun shy. But could be I just need time to recuperate from that and to feel I'm making progress again by focusing on "my own thing" for awhile. I might be somewhere on the autism spectrum myself: sometimes interacting with people is exhausting for me.)
Robin
Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Feb 26, 2:35 am, snipped-for-privacy@gmail.com wrote:

Robin,
All good points. This hobby does seem to snowball and take up a lot more resources than one expects.
There is a popular servo erector set called Bioloid. I'll see if I can come up with a proof of concept using this set. These are smart servos that can be synchronized with high speed communications (1Mbps).
I'm thinking of two key approaches - making the "keepon" even smaller because less travel means faster looking responses to motion commands. Also, the camera will be a cheap usb camera with mic that is actually hidden in the base. To simplify, the camera will not move but I think face tracking software should still work as long as a person is in the field of view.
Danh
Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I'd heard of Bioloid but this is the first time I've looked at it. Looks nice!

Interesting. It will also be interesting to see the psychological effect of that. (Is smaller cuter? Or the reverse?) My first attempt may end up being larger, simply because I don't want a size constraint to keep me from succeeding. If it works at all, I'll want to scale it down though.

Oh yeah, I keep forgetting these things have a mic in them too! Good idea :)

Umm. There I'm not sure. I presume you'll measure the fov cone and compare to the range of motion to see if people do remain in view.... Users will want to interact with the robot's "face" so they may present at an angle that's hard to detect. (Profile is hard, and so are under-the-chin views -- too few reliable features.) If you'd like to capture some worst-case test images before you start to build, I can evaluate feasibility.

I'll really look forward to seeing what you come up with. I'll post mine as I go -- be good for a laugh :)
Robin
Add pictures here
✖
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.