Re: New worlds

Olivier Michel <[email protected]>
Newsgroups gmane.comp.cybernetics.webot.contest
Organization Cyberbotics Ltd.
Message-ID <[email protected]>
Hi,

Thanks for all the feedback. It seems that although the new worlds are 
breaking most behaviors of existing robots, adapting the robots to these 
new conditions is not impossible. Please read my comments at the end of 
this message.

Mike wrote:

>The new world is more challenging, but not too restrictive.
>
>I was able to modify my robot controller to stand up from a fall in the new
>world within the 5 second time limit.
>I have not completed modifying my robot's walking and turning routines yet,
>but I think they only need a few modifications to keep the robot from
>slipping.
>  
>
Thanks for these investigations. It seems that the robots slips more in 
this new world ? However, do not fine tune everything right now since we 
may still modify a little bit these worlds (see below).

Benoît wrote:

> gollum is not a very good robot, but here what it gives with the new 
> world : * it is walking, but a bit more slowly (it is like sliping on 
> the floor) * It is enable to turn on one leg, but not at the same 
> angle as the one given to the leg (sliping again) * it can no longer 
> get up after a fall I tried a but of GA, it seems to work, but after 
> more than 150 resets, I got (after 5 minutes) 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=-0.0000e+00) ODE Message 3: LCP internal error, s <= 0 
> (s=-1.7077e+07) ODE Message 3: LCP internal error, s <= 0 
> (s=0.0000e+00) and the robots disapered from the world. the gps matrix 
> seemes full of NaN With the previous world, webots was able to run for 
> more than 24h and I have never get any error. What I think about this 
> world : except the ODE Message bug, it seems good but I need a bit 
> more experimentations 


Thanks for this feedback. Although I am not sure why you get those LCP 
internal errors, it might come from the fact that the robot is much more 
heavy now... I maybe should also increase the friction coefficients to 
fix this issue and avoid the slippery.

Carmine wrote:

> Oliver,
>     I did a quick and rough tests. My robot is no longer able to walk, 
> rotate and get up. :P
> I need to redisign all the procedures even because I do not know 
> anything about genetic algorithms. I hardcoded my movements empirically!
> For me it is not a problem, but I would like to be sure we are going 
> to use that new world before making any effort in redeveloping the code.


Currently, this new world is just submitted for review. I will let you 
know if we decide to adopt it. However, I think it will change a little 
bit before we adopt it.

Tsuy wrote:

>hi olivier
>
>I found the new judo world not optimal for my controller (Gogo_Yubari). The
>idea with the selfcollision i found a good idea. but why do we change the
>angles range? is this necessary to avoid selfcollision or is this sony's
>robot-specification?
>
>tsuy
>  
>
Sony's robot-specification are unkown, but for sure in any robot, servo 
angles are limited.  So I tryed to set decent values for each servo 
(although they do not match exactly any existing robot). Moreover 
self-collision cannot be applied to two consecutive joints and joints 
limits should be used to avoid self-collision with consecutive joints. 
So, I believe that servo limits are necessary. However, I am open to 
discuss every limit value and fix those that are causing troubles. For 
example the hip_3 limits may prevent some robots to perform a beautiful 
"splits" down to the floor. If you like we might decide to increase 
these limits, so that the "splits" can reach the floor. But is it really 
necessary ?

I attached to this message a modified judo.wbt world with greatly (x10) 
increased friction coefficients (coulombFriction and 
forceDependentSlip). This new world should behave more like the current 
official world concerning friction issues (and possibly LCP solver as 
well). I appreciate your feedback about it.

-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/
judo.zip (application/zip, 18.6 KB) - not displayed
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.