Home Creators Posts Import Register Favorites Logout
haven't archived this post yet. have a subscription? use the importer!

Content

The electronics are now installed and I've been able to power up the motors. I'm still pretty confident this will work ok, although I could probably have done with a higher belt reduction ratio all over, but I'm pretty confident it'll walk better than anything I've built to date.

The legs seem pretty springy and they move really fast, so it'll probably be able to 'prong' which is jumping along on all four legs at once. There appears to be enough force for this also.

Reading force on each joint is of course easy because I can read the actual encoder position vs the demand position to get a force reading (like measuring stretching a spring), so that will allow for a lot of dynamic control. This will follow on from the R&D I did last year ( https://www.youtube.com/playlist?list=PLpwJoq86vov9Aogc6PseW8BUe2nQyZESZ ).

Full video will be up for Patrons early next week!



Files

Previews only

Comments

Scott Cress

I am super excited about this. Need to set up my 3d printer so I can build one. 😄

Anonymous

Really awesome! I'm curious on how far you can take this before another hardware iteration is needed. Do you plan on focussing more on the Software side of things once you found a good dog HW solution?

Anonymous

Hype hype hype!!!!! As jumpysquirrel mentioned, once you have the hardware working well is it time for software? I think this would be a great platform to try some AI on. You could have it learn to balance through trial and error (Q-learning) and perhaps optimize its different base walking gaits for speed as well. A simulation of the physical design would be nice too, as you could train the simulation and try and transfer that experience to the robot. Regardless.. I can feel the hype :-)

Anonymous

Nice! I really want to get into robotics again after I finish my current software project.

XRobots

There will definitely be hardware upgrades to this one - and it may be that every piece gets replaced in the eventual process, but I don't anticipate a radically different design any time soon, so it will more of a gradual process if it changes.

XRobots

These are all things I've been wanting to try for a while. I plan to learn ROS also, and even use that on wheeled robots if openDog isn't ready.

Anonymous

Was this test at 24V?

Anonymous

I noted with interest your point about the motors getting warm, and the PLA possibly not coping. Have you considering doing a silicone mold on the part, and pouring a higher temperature resin? Or would that be less ideal than a cnc'ed part? I'm not sure on the performance/cost benefits between the two.

XRobots

I'll be talking about that in the next video - I think we ultimately need an aluminium plate/heatsink fitted in there for the long term. Possibly 3D printed PETG or Nylon might do also. The other option is carbon fibre sheet which is bonded with epoxy because it doesn't melt as far as I understand it.