The robot playing has been proceeding, if slowly, over the past wk/10 days.

It’s a pity the same can’t be said for those guys at Youtube. Every time I’ve tried to use it since before the Easter break it’s been "the site is undergoing maintainance; that feature isn’t available". Including uploads.

So it looks like I’ll have to investigate moving to alternative sites — plenty of them available, I’m sure. My problem will be the volume. Already, it seems, my 80-100 vids seems to put me in the top .1% of YouTube clients.

A similar thing happened when I was using BlogSpot a couple years back. They never envisioned a guy that could (or would WANT to! type 10,000 lines a day, every day. Their idea was more along the 100-1000 line region maybe a couple times a wk.

At least I managed to force (in the course of 4 or 5 years) them to fix their initially-underwealming platform.

I’ll add the clips of the latest "humanoid" vids when I manage to put them up somewhere.

I’m at the point where the Robotis s/w simulates OK, but the humanoid in VR is showing all the problems of falling over the IRL version shows on a table that isn’t 100% flat and unyieldingly hard. There’s also (as usual) some problem with the frictional force calcs so the feet slip in the right places — some of the poses require the feet slip so the bot can move its legs without having to lift its feet — but also seem to entail slipping in unexpected places (e.g. when the humanoid tries to turn its upper body, all that happens is the lower body rotates and the bot ends up facing the same direction +- 1 degree).

Most of the problems relate to the new feature (for my simulation project) of feet that are extended plates rather than points. The "stickbot", "myquad" and "king spider" have all made contact with the floor at N points (for sure some of the feet were "brackets" so the points in question were typically corners of some bracket).

The old standby of having the s/w automatically select several points on the bottom of the plate-feet and just "call" them "point feet" so the old s/w can work just isn’t good enough for the humanoid. That ploy worked OK for the stickbot standing and balancing on two or one foot, but just results in silly behaviour here.

As usual, the bugs will result in a hail of patching to low-level code before I can figure out how to elegantly state the required new constraints as "never" and "always" rules in the appropriate top-level planner module. (Such modules are associated with the robot assembler as well as different stages of the motion planning and rendering process, and in addition several "add ons" — mostly related to "number of feet this robot has" — are presently required for each diffrent class of Bioloid you’re trying to simulate).

OK. I’ve uploaded the vids to Google videos. But — of course — they don’t display when added to the std Drupal video links.

So here’s one explicit URL: http://video.google.com.au/videoplay?docid=806932421774407010&hl=en-AU