feeback answers

<[email protected]> wrote:

>Hi
>
>I have tried the new worlds with Kubiak's controller. The robot is not able to do anything ( walk / turn / get up... ) anymore but it should not be very difficult to adapt him to the new conditions. I think the joint limits will improve the quality of the contest and I agree with Mike to say that the new friction coefficients are better: it seems important to me to avoid slipping as much as possible. Just one question :  I suppose real robots can't do a 360 degrees with their head, so why is there no joint limit for the neck?
>  
>
I think there is a limit althought I forgot to mention it in the e-mail, 
but I believe there is also a bug in Webots allowing you to pass through 
unreachable position to reach a legal position... I have to fix this.

Mike Jost wrote:

>My robot was not able to walk forward at all with the original friction
>coefficients due to slipping.
>With the new friction coefficients, my robot is able to walk and turn.
>  
>
Thanks!

Marc Breuer, Sonovabic wrote:

>I've just been testing the new world with Sonovabic,
>none of his moves work any more (at the moment). I've
>had a good look at the restrictions, they seem to be
>quite human-orientated. for a robot however I don't
>think there should be the same joint-limits (unless of
>course you would like to limit to humanoid-robots).
>
>During the design of Sonovabic I made pretty much sure
>not to create unlikely movements caused by
>intersections and it should be possible to design a
>robot capable of the tasks he does. I would be glad if
>it could be possible to limit the angles to
>robot-concerned limits and not to human ones. for
>example it shouldn't be difficult for a robot to bend
>the elbow backwards (not that I use that).
>
>In my opinion it would be sufficient to limit most
>angles to something like this -3.0 -> +3.0 , as it
>would avoid most self intersections. the neck and
>shoulder_1 should be the only unlimited ones. Even
>with the new settings it will be possible to cause
>intersections, especially joints with 3 dimensions of
>movement are hardly controllable. I don't think it can
>be really managed by limiting joints.
>
>Another solution would be to allow a sort of
>angle-offset. e.g. the elbow joint has an angle limit
>of 0-2.8 , adding an offset of 0.5 would result in
>angle-limits of 0.5-3.3. these values would have to be
>set during initialisation and may not be altered
>during a match to avoid misuse. the offset itself
>could also be limited to avoid intersections.
>
>I wouldn't be too pleased at a change now, as it would
>render all my work so far obsolete.
>
I am afraid we would like to head towards human like behavior. This is 
mainly motivated by the fact that most humanoid robots implements joints 
limits similar to human movement limits. Humanoid robots are not only 
humanoid by the shape and structure, but also by the joints limits. And 
as this contest is humanoid robot centered, we will try to keep acurate 
with real humanoid robots.

> the increase of
>friction does make sense as slipping is certainly a
>problem.
>
>As we are talking about new settings, I would like to
>see a new neck-servo enabling looking up and down, and
>a rotation of the upper body 
>  
>
Good idea! I will add it to the model if everyone agrees.

Benoît wrote:

> gollum is no longer able to do side steps the way it did before: Just 
> bending is one direction and then changing hips angles. But maybe this 
> is the way it should be in "real life". Is is now slipping only a bit 
> (no perfect 90 degrees turns), again, I think than this is how it 
> should be : there is a bit more inertia. After some GA computation, I 
> got very fast the LCP problem again. gollum is able to walk anyway. 

>With the new world, I got 
>
>ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)
>ODE Message 3: LCP internal error, s <= 0 (s=-0.0000e+00)
>ODE Message 3: LCP internal error, s <= 0 (s=nan)
>ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)
>
>after 15 minutes
>
>sorry...
>
This LCP problem could be tackled out by adjusting some physics 
simulation parameters in ODE, like ERP or CFM. The next version of 
Webots will allow you to tune these parameters to see if it helps...

-Olivier

 

------------------------ Yahoo! Groups Sponsor ---------------------~-->
Buy Ink Cartridges or Refill Kits for your HP, Epson, Canon or Lexmark
Printer at MyInks.com. Free s/h on orders $50 or more to the US & Canada.
http://www.c1tracking.com/l.asp?cid=5511
http://us.click.yahoo.com/mOAaAA/3exGAA/qnsNAA/pU_rlB/TM
---------------------------------------------------------------------~->

Yahoo! Groups Links

To visit your group on the web, go to:
 http://groups.yahoo.com/group/webots-contest/

To unsubscribe from this group, send an email to:
 [email protected]

Your use of Yahoo! Groups is subject to:
 http://docs.yahoo.com/info/terms/