2 more DOF on the hips AND stock moves?

[limor]

Great work
Any chance you can post the rigged RN1 model?

[Gil]

Good idea. I wonder how it will behave in the real world

[hivemind]

well, as seen in the picture(below), he would obviously kick ass completely...

now that i have tweaked a few other thing, rigged the model and added weight for physics i think hes doing pretty well. you may notice his right foot (directly under him) looks funny, this is simply because his foot ratio is so large, but i am a black belt in karate, and i know what i flying side kick is supposed to look like:



and you may also think that the head cannot rotate like that (as it cannot stock) but thats the miracle of a small steel plate and another servo...

[pepperm]

Very nice picture Hivemind. Being new to this are of robotics I'm fascinated by the picture. How did you produce it? I'd love to play around with some pictures like that whilst I wait for my humanoid to be born.

Regards

Mark

[limor]

added weight for physics

are you running it through a simulator? what system are you using?
8O

[hivemind]

im working on putting my model in a phyics engine, but i only started my model about 3 days ago and im not positive how im going to implement the whole phyics idea. I was thinking about using havok reacter (as i already have a good idea how to use that) and have my model as a rigid ragdoll thing (which i would probably place on a stand) so you would eventually be able to manipulate the servos and use the numbers (a copy/paste system similar to robobasic). but i have work that im busy with and ive never run an interactive largescale physics sim. so we'll see where it goes.

i have plans to create the sim, but i dont know how good it will be for others seeing as my rn1 will be far from stock.. perhaps a kind of adaptive model would be best, having a user ability to control which servos they have on and off, whether they have grippers, and if they do, which motor ports they are on. but if so then I am a ways away.

in the mean time i have a completely functional model that seems to have most of the bugs worked out, unfortunetely, due to my new legs, it wont be able to use stock moves, althought straight walking ought to be very similar to that of a real person do to the knee and upper leg movements. i have some 'perfect' human walking gaits on my computer, so once i write a script to capture the angular movement of those joints i should be able to implement it to a gait on my model...

[limor]

There are several simulators that mix 3D mesh with physically simulated objects. I wrote an opensource one called ezphysics.org ( you can download a demo http://ezphysics.org/EZPhysics-19Oct2005.rar ) and the last couple of weeks Microsoft have nonchalantly announced their physics simulator.

If you post a low-mesh-count version of your the rigged model, i can have it ragdolled in no time with ezphysics. you can then access the physically-rigged model programatically in order to simulate the servos and controller board..

[hivemind]

yea, my meshes are still pretty high poly (for the renderings i like things to stay smooth, its just the kind of guy i am...) and im starting to work on my own physics simulation, so hopefully i can get that running.

perhaps its time to start a new thread dedicated to making a model/physics engine to see if there are other that can help or at least have ideas how to make it better and easier to use. id like to get to the point where i can simulate a program right inside the 3d engine having the whole thing work like one normally would, im just not sure what the computing requirements for something like that may be.

[H3dude]

If you post a low-mesh-count version of your the rigged model, i can have it ragdolled in no time with ezphysics. you can then access the physically-rigged model programatically in order to simulate the servos and controller board..

Limor, I can reduce the polycount and make a simple mesh for you if you like, and if it's ok with hivemind. I haven't looked at the files yet, but I am very adept at low-poly modeling for Havok and MathEngine. I assume you don't want any screws/bolts or small items?

[limor]

H3dude, thanks for helping. most systems can deal with 100k's of poly at 50 fps so i wouldnt worry about throwing out the screws/bolts. We can test a couple of models and see how it goes. As long as it has bones at the right servo axes we should be good to go.

[hivemind]

My machine has 2 gigs ram, 3.2 gig dual processor and some geforce card, 6600 i believe, regardless, i have been playing with my model in havok and it seems to be working fine, but for whatever reason it seems to have some small bugs one of the leg meshes... probably a stray vertex somewhere... who knows, but im working on figuring it out

as for now ill just post a small walking vid. it looks pretty similar to the stock walking - jerky and such- but its a start to make sure the model works... we'll see

RN-1 Walking

[robodude666]

That is mighty cool. I checked out ezphysics today. It looks pretty cool. The editor is like webots, but a lot simpler and easier to use. controls in 3d view suck. try importing or whatnoting your model into there and see if it works.

[H3dude]

guys, I know that Havok will work well even with such a high poly model. The collision capsules you set in Havok's editor are all that matter. I still think the screws and bolts will be superfluous for running a physics simulation but that's just my 2 cents. If you want me to help out please let me know. I'd love to be involved.

[CaptKill4Fun]

ok, so my RN-1 is on the operating table awaiting gears after a nasty fall prompting me to work on my digital RN..

It's not to spec and not extreme detail, but Looks About Right so far, plus ive also had some thoughts on adding that sought after extra rotation to the leg..

Hi,

Nice job ... and the twising hip joint is almost inline with a design I'm working out! So far it will be perhaps 2mm to 4mm between the lower body brackets and upper leg brackets ...

I researched the best low-profile mounting system to fit between the servos and the existing brackets ...

I designed a new bracket to replace both brackets and keep the vertial dimensions much closer to the original RN-1 and other robots ...


I also make several new parts for other robots too ...

[shsan]

Hi Hivemind,

Really nice job on the model.

I have been working in adding the KHR1 to the Microsoft Robotics Simulation and it's almost done.

The mesh for the simulation is way simpler as with the havok engine. In my case i just put some bounding boxes. (Might need to be refined though).

I would be interested in having your complete model also to add it too if you don't mind.

Or the opposite if you want you take my code and patch it to match the RN1

Don't hesitate to contact me directly if you want.