Re: webots contest

Hi,

Joel Chestnutt wrote:

>Xavier Bitot wrote:
>  
>
>>1)      I haven't noticed any change at all in my robot between webots 4 and 5.
>>Are the physics of the game definitive until the end of the
>>contest or is there a chance that they change before May 1st? You
>>(Olivier) have been talking a few months ago about a self-collision
>>test but it would probably break many robot moves….  Since there
>>are now less than two months to go, in my opinion it is time to
>>"freeze" the physics so that everybody can focus on building moves
>>that will not have to be changed.
>>    
>>
>
>I think self-collisions would be a good addition, but I agree that if 
>it's going to be included, it should be done soon and then the physics 
>kept the same.  Also, if any changes like this are going to happen to 
>the physics, it would be nice to have some warning so we could start 
>adjusting our robots now.
>  
>
I wont be able to modify anything until the end of the contest as I am 
extremely busy at the moment. So, don't worry, the rules won't change. 
We are planning to improve the contest with self collision detection 
only for next contest.

>>2)      I have been following the evolution of the hall of fame since last
>>summer and it is very interesting to notice that it has been changing
>>a lot all the time even when the robots were hardly ever updated.
>>    
>>
>
>I don't know how much I can trust the "last modified" column in the hall 
>of fame.  From what I can tell, it checks the modified date of the main 
>robot class.  I updated toddler last week, but because I was only 
>changing some of the motions and other classes, it didn't change 
>toddler's "last modified" date.  I don't expect the 
>ToddlerInSnowsuit.class file to need to change at all for the rest of 
>the contest, in fact, as the important things are handled in other 
>classes.  However, I don't know how common this is with the other 
>competitors, and may just be unique to toddler's code organization.
>  
>
I tryed to fix the "last updated" column of the hall of fame. Now, it is 
a bit better (it looks at all the files in the robot's directory and not 
only the *.class files as it used to do), but not perfect (as it doesn't 
yet go recursively into sub directories to perform this check).

>>3)      Webots doesn't react exactly the same on all platforms and the
>>possibility of playing off-contest matches is a great improvement
>>compared to last year, it gives us the opportunity to try our
>>robot moves on the server without having to wait for a day, I really
>>like it. At the moment only one set per match is run in off-contest
>>games. Would that be possible to play for example a 5 sets match
>>instead, so that we can download several sets at once and try
>>different things according to the score without having to upload a new
>>controller after each set or use a random strategy choice?
>>    
>>
>
>I agree, but I still think running quick matches to test little things 
>is very useful.  I'd like a drop-down box on the test match page to 
>choose how many rounds to do.
>  
>
About running several off-contest matches, I'd prefer not to change this 
because there are sometimes several competitors attempting to run a 
match a the same time and this is not possible. Running several matches 
in a row would make it more difficult for competitors to get the 
opportunity to start a match. Moreover it would take some time to 
program this...

>>5)      Is it not possible anymore to upload a zip file on the server?
>>I tried it yesterday afternoon and it didn't work, but uploading the files
>>one by one was fine.
>>    
>>
Could you check that your zip file is not corrupted. Normally, it works.

>I've only ever uploaded zip files for toddler.  If that's not working 
>anymore, it's definitely going to be a problem.  Toddler's code is 
>organized into packages, which means a directory structure is necessary. 
>  I just create a zip with all the necessary directory structure, and 
>upload that.  The server unpacks it, preserving the directory structure, 
>and everything works nicely.  It was working as recently as last week.
>  
>

>I have a couple of other things I'd like to ask or mention:
>
>- The LED settings for eyes and such don't seem to get recorded into the 
>wva files, or if they do, it's not getting displayed.  It would be nice 
>if we could get that to help debug what the robot was thinking when it 
>does something stupid in a match.
>  
>
Sure. I will add this on my TODO list...

>- The "judo" characters on the mat are backwards when displayed in the 
>windows webview program.  They're fine in the simulator and in the linux 
>webview, but for some reason, they're flipped in the windows webview. 
>Yeah, I know, this is pretty minor.
>  
>
This is already fixed in my development version. However, since it is 
minor, I won't release a new version just for that.

>- Would it be difficult to change the "last modified" column to reflect 
>the last time anything was uploaded to that robot?
>  
>
Yes indeed. I tryed, but this is not perfect. I will try again...

>- How exactly does the distance sensor work in the simulation?  It seems 
>to vary wildly during test matches.  I've been assuming that it is the 
>distance to whatever is at the center pixel in the camera's image.  Is 
>that true?
>  
>
No. You should go to the Preferences of Webots and check the box that 
display the range of the distance sensor, then you will understand that 
it is not necessarily the center of the image.

>- The legs are mirror images in configuration, but that's not how it 
>appears in the diagram here: http://roboka.org/images/judoka_servos.png
>The hip servo order doesn't match between the left and right hips.  I 
>believe the right hip order is the correct one.
>  
>
Do you mean that some rotation arrows need to be fixed on the image ? 
Actually I am not the author of it, I believe it was contributed by 
Frans Klip <[email protected]>. I would be glad to publish a 
fixed version of it if anyone can contribute it.

>- How noisy is the GPS measurement in the online matches?  It's 
>perfectly accurate offline.  We've started using it for some things 
>recently, which seem to work fine online as well.  The documentation 
>says the the position measurements are noisy online, but I haven't seem 
>it seeming to affect anything in the online matches.
>  
>
Actually, it is not noisy online either. We never implemented this noise 
for contest matches... Shall we do it before the end of the contest ?

-Olivier


------------------------ Yahoo! Groups Sponsor --------------------~--> 
Has someone you know been affected by illness or disease?
Network for Good is THE place to support health awareness efforts!
http://us.click.yahoo.com/RzSHvD/UOnJAA/79vVAA/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/