simspark-devel — Discussion about Spark development

Re: [simspark-devel] new problem of new server

[Joschka B. <jbo...@un...>]

Hi Yuan,

On Sat, 2007-06-16 at 02:42 +0800, Yuan Xu wrote:
> Hi Joschka,
> 
> I do some tests under the situation that one agent connected server, I
> faced new problems ;-(

Which version are you using, current CVS or the 0.5.6-preview?

> (1) see http://picasaweb.google.com/xuyuan.cn/RCS/photo#5076361988020790178
>     * Situation: after beam action
>     * Problem: the lower torso rotation is not correct

Does this happen everytime? I have never observed this as of yet...

> (2) see http://picasaweb.google.com/xuyuan.cn/RCS/photo#5076361988020790194
>     * Situation: after first joint action, the action was "(rae1_2
> -0.137532 0.623955)"
>     * Problem: the body disjoint while the agent starting use joint
> effector, and after that moment it seems OK.
>                However this is very bad, because when the agent try to
> walk, it will jump at the beginning ;-(

Could you please try to comment out the joint stops in the soccerbot.rsg
file? I suspect them to have something to do with this problem. 

The joint stops were easy to implement, but I'm afraid that some ODE
parameters might have to be adjusted to be able to use them in a stable
way. If this is the case, I have to remove the joint stops again for
now, since we don't have much time to find good parameters currently :-(

> I hope we could fix them quickly.

Me too, believe me ;-) Thanks for your help!

Joschka

Re: [simspark-devel] doubt about FRP

[Hedayat V. <hed...@ai...>]

Hi,
Thank you for your feedback on FRP. When the foot does not touch any 
thing, FRP should not be sensed. In my own tests (normal cases, not 
flying!), it is the case. FRP is sensed only when a foot is on the 
ground. I've not tested it, but if the agent is flying, and his feet 
doesn't touch anything (even it's own body), FRP should not be sensed.
I've checked my code again, and I think that if there is any FRP sense, 
so certainly ODE has reported at least one collision. Please check it a 
little more to see if FRP is sensed when there is no collision and let 
me know the results.
Hopefully, it should be OK. And the output you've provided could be 
really correct! As the output shows, all the applied force are from sides.

Thanks a lot,
Hedayat

/*"Yuan Xu" <xuy...@gm...>*/ wrote on 06/15/2007 08:35:08 PM:
> Hi Joschka and Hedayat,
>
> I have a doubt about Force Resistance Perceptor:
> It provides information about collision,
> however in my test, the agent receive such perception:
> (FRP (n lf) (c 0.04 0.12 -0.05)(f 12.61 10.39 0))(FRP (n rf) (c -0.48
> -0.02 -0.06)(f -11.01 17.48 0))
> while the agent is 'flying'!
> In my opinion, if the foot does not touch any other thing, the force
> should be zero or the FRP should not be sensed.
>

[simspark-devel] new problem of new server

[Yuan X. <xuy...@gm...>]

Hi Joschka,

I do some tests under the situation that one agent connected server, I
faced new problems ;-(

(1) see http://picasaweb.google.com/xuyuan.cn/RCS/photo#5076361988020790178
    * Situation: after beam action
    * Problem: the lower torso rotation is not correct

(2) see http://picasaweb.google.com/xuyuan.cn/RCS/photo#5076361988020790194
    * Situation: after first joint action, the action was "(rae1_2
-0.137532 0.623955)"
    * Problem: the body disjoint while the agent starting use joint
effector, and after that moment it seems OK.
               However this is very bad, because when the agent try to
walk, it will jump at the beginning ;-(

I hope we could fix them quickly.

-- 
Best wishes!

Xu Yuan
School of Automation
Southeast University, Nanjing, China

mail:   xuy...@gm...
          xy...@ya...
web:   http://xuyuan.cn.googlepages.com
--------------------------------------------------

[simspark-devel] doubt about FRP

[Yuan X. <xuy...@gm...>]

Hi Joschka and Hedayat,

I have a doubt about Force Resistance Perceptor:
It provides information about collision,
however in my test, the agent receive such perception:
(FRP (n lf) (c 0.04 0.12 -0.05)(f 12.61 10.39 0))(FRP (n rf) (c -0.48
-0.02 -0.06)(f -11.01 17.48 0))
while the agent is 'flying'!
In my opinion, if the foot does not touch any other thing, the force
should be zero or the FRP should not be sensed.

-- 
Best wishes!

Xu Yuan
School of Automation
Southeast University, Nanjing, China

mail:   xuy...@gm...
          xy...@ya...
web:   http://xuyuan.cn.googlepages.com
--------------------------------------------------

Re: [simspark-devel] About SoccerRules

[Joschka B. <jbo...@un...>]

Hey Markus,

On Mon, 2007-06-11 at 19:30 +0200, Markus Rollmann wrote:
> Markus Rollmann wrote:
> [...]
> > yes, I think the problem is that in the recursive case the result of
> > FindChildSupportingClass(...) for the first found child is immediately
> > returned. So if the first child does not have a matching node the function
> > returns immediately false instead of testing further children.
> > 
> > The fix should be to add a second inner loop that checks all children of the
> > current node and then returns for the first found node.
> 
> I commited a fix to the simspark CVS, a second loop wasn't necessary. 
> Find the diff below.

Great, thanks a lot!

cu
Joschka

Re: [simspark-devel] About SoccerRules

[Markus R. <rol...@un...>]

Markus Rollmann wrote:
[...]
> yes, I think the problem is that in the recursive case the result of
> FindChildSupportingClass(...) for the first found child is immediately
> returned. So if the first child does not have a matching node the function
> returns immediately false instead of testing further children.
> 
> The fix should be to add a second inner loop that checks all children of the
> current node and then returns for the first found node.

I commited a fix to the simspark CVS, a second loop wasn't necessary. 
Find the diff below.

regards,
Markus

Index: leaf.h
===================================================================
RCS file: /cvsroot/simspark/simspark/spark/zeitgeist/leaf.h,v
retrieving revision 1.3
retrieving revision 1.3.2.1
diff -C2 -d -r1.3 -r1.3.2.1
*** leaf.h	17 May 2007 17:46:45 -0000	1.3
--- leaf.h	11 Jun 2007 17:27:05 -0000	1.3.2.1
***************
*** 130,134 ****
                   if (recursive)
                       {
!                         return 
(*i)->FindChildSupportingClass<CLASS>(recursive);
                       }
               }
--- 130,138 ----
                   if (recursive)
                       {
!                         child = 
(*i)->FindChildSupportingClass<CLASS>(recursive);
!                         if (child.get() != 0)
!                         {
!                             return child;
!                         }
                       }
               }

Re: [simspark-devel] About SoccerRules

[Markus R. <rol...@un...>]

Hi,

On Sun, 10 Jun 2007, Joschka Boedecker wrote:
[...]
> This is not the only place where FindChildSupportingClass is used, and 
> in other cases, it seems to work (as does ListChildrenSupportingClass 
> which basically works the same way). Maybe it has something to do with 
> the recursion, but from looking at the code in lib/zeitgeist/leaf.h, I 
> couldn't find anything suggesting that. Markus, in case you read that, 
> do you have any idea why this could happen?

yes, I think the problem is that in the recursive case the result of
FindChildSupportingClass(...) for the first found child is immediately
returned. So if the first child does not have a matching node the function
returns immediately false instead of testing further children.

The fix should be to add a second inner loop that checks all children of the
current node and then returns for the first found node.

Cheers,
Markus

Re: [simspark-devel] Should I add the FRP perceptor to the agent?

[Joschka B. <jbo...@un...>]

Hi Hedayat,

hed...@ai... wrote:
> Hi!
> As you've noticed, I've not changed any rsg files to enable the FRP 
> perceptor as I was not sure that you want to include it or not. Is it 
> going to be used in Atlanta?

Personally, I would say it is preferable over the simple binary touch 
sensor. The last call on this is up to the organizers though. Feedback 
from the teams would make this decision easy, but there hardly is any 
feedback on anything but bugs in the server :-( I'll probably write 
another mail to the ML, asking teams to tell us their experience with 
different features like the FRP, the timer, the modified SoccerBot, and 
the proposed ball model.

I'm really starting to get frustrated by the lack of cooperation...many 
people just don't seem to care :-( (those who do care do a great job 
though :-) )

Cheers,
Joschka

[simspark-devel] Should I add the FRP perceptor to the agent?

[<hed...@ai...>]

Hi!
As you've noticed, I've not changed any rsg files to enable the FRP perceptor as I was not sure that you want to include it or not. Is it going to be used in Atlanta?

Cheers,
Hedayat


________________________________________________________________________
Check Out the new free AIM(R) Mail -- 2 GB of storage and industry-leading spam and email virus protection.

Re: [simspark-devel] About SoccerRules

[Joschka B. <jbo...@un...>]

Hi Hedayat,

hed...@ai... wrote:

> Thank you for your help. It'll really save my time. (I would probably 
> missed the GetWorldBoundingBox function!).

:-)

> A question (which will help me to understand the code better):
> The original ClearPlayers functions were using GetAgentBody function 
> which uses FindChildSupportingClass function. It was unable to find 
> any Body node in the AgentAspect's children (body AgentAspect).

Yes, I also noticed that. I changed the GetAgentBody method back to use 
the GetChildOfClass function. It should work now.

> But when I modified the GetAgentBody function to use 
> GetChildSupportingClass (or maybe the GetChildOfClass, I don't 
> remember now!), it successfully returned the first Body object. (This 
> is the function that the GetBody function uses). Why the 
> FindChildSupportingClass function is unable to find an object of the 
> Body class while there is such an object in it's children?!

Exactly this problem caused me lots of headaches yesterday, and I 
finally gave up to try and understand it. The Body child node is 
definitely there in the hierarchy, and it is easily found by 
GetChildOfClass, but the FindChildSupportingClass (with recursion 
enabled) will miss it. The FindChildSupportingClass method tries to 
dynamically cast the shared pointers to the nodes it finds when 
descending the SceneGraph, and to my understanding, a cast from a Body 
node to a Body node should be successful. It seems that the casted 
pointer will always be NULL though :-(

This is not the only place where FindChildSupportingClass is used, and 
in other cases, it seems to work (as does ListChildrenSupportingClass 
which basically works the same way). Maybe it has something to do with 
the recursion, but from looking at the code in lib/zeitgeist/leaf.h, I 
couldn't find anything suggesting that. Markus, in case you read that, 
do you have any idea why this could happen?

Cheers,
Joschka

Re: [simspark-devel] About SoccerRules

[<hed...@ai...>]

 Hi,
Thank you for your help. It'll really save my time. (I would probably missed the GetWorldBoundingBox function!).
A question (which will help me to understand the code better): 
The original ClearPlayers functions were using GetAgentBody function which uses FindChildSupportingClass function. It was unable to find any Body node in the AgentAspect's children (body AgentAspect). But when I modified the GetAgentBody function to use GetChildSupportingClass (or maybe the GetChildOfClass, I don't remember now!), it successfully returned the first Body object. (This is the function that the GetBody function uses). Why the FindChildSupportingClass function is unable to find an object of the Body class while there is such an object in it's children?!

Thanks for your patience,
Hedayat


-----Original Message-----
From: Joschka Boedecker <jbo...@un...>
To: Hedayat Vatankhah <hed...@ai...>
Cc: Simspark Devel ML <sim...@li...>
Sent: Sun, 10 Jun 2007 11:11 am
Subject: Re: [simspark-devel] About SoccerRules









Hi Hedayat,?
?

Hedayat Vatankhah wrote:?

> Hi Hesham,?

> OK, from now, I'll (hopefully!) add new function(s) to calculate the 
> AABB2 or AABB3 around an agent to be used instead of a single position 
> in functions such as ClearPlayers functions.?
?

Just some thoughts which might make this task easier:?
?

Each agent is linked into the SceneGraph at a position 
"/usr/scene/AgentAspectX" in the virtual file system of the zeitgeist 
framework, with X being the ID of the agent. All the nodes of the agent 
will be below this node. The actual AgentAspect of the SoccerBot, e.g. 
for agent with ID 1, will then be at:?
?

/usr/scene/AgentAspect1/spaceman/body?
?

The "spaceman" is the collision space that encapsulates all the nodes of 
the agent, and "body" is the name that we set for the AgentAspect in the 
soccerbot.rsg file.?
?

If you have the AgentAspect of a certain agent and you want to get the 
bounding box containing all of the agents bodies, it should be 
sufficient to get to the collision space using GetParent and then using 
the GetWorldBoundingBox method (implemented in the BaseNode class, from 
which Space is derived). Just to be sure, you could do a UpdateHierarchy 
(also from BaseNode) before you get the bounding box, but I think it is 
not necessary.?
?

Cheers,?

Joschka?



 


________________________________________________________________________
Check Out the new free AIM(R) Mail -- 2 GB of storage and industry-leading spam and email virus protection.

Re: [simspark-devel] About SoccerRules

[Joschka B. <jbo...@un...>]

Hi Hedayat,

Hedayat Vatankhah wrote:
> Hi Hesham,
> OK, from now, I'll (hopefully!) add new function(s) to calculate the 
> AABB2 or AABB3 around an agent to be used instead of a single position 
> in functions such as ClearPlayers functions.

Just some thoughts which might make this task easier:

Each agent is linked into the SceneGraph at a position 
"/usr/scene/AgentAspectX" in the virtual file system of the zeitgeist 
framework, with X being the ID of the agent. All the nodes of the agent 
will be below this node. The actual AgentAspect of the SoccerBot, e.g. 
for agent with ID 1, will then be at:

/usr/scene/AgentAspect1/spaceman/body

The "spaceman" is the collision space that encapsulates all the nodes of 
the agent, and "body" is the name that we set for the AgentAspect in the 
soccerbot.rsg file.

If you have the AgentAspect of a certain agent and you want to get the 
bounding box containing all of the agents bodies, it should be 
sufficient to get to the collision space using GetParent and then using 
the GetWorldBoundingBox method (implemented in the BaseNode class, from 
which Space is derived). Just to be sure, you could do a UpdateHierarchy 
(also from BaseNode) before you get the bounding box, but I think it is 
not necessary.

Cheers,
Joschka

Re: [simspark-devel] About SoccerRules

[Joschka B. <jbo...@un...>]

Hi Hedayat,

Hedayat Vatankhah wrote:
>
> I've changed MoveAgent for the new agent (thanks BeamEffector!) and 
> fixed ClearPlayers functions.
>

Great, thank you. I've just committed a bunch of changes to the 
SoccerBase (mainly to increase performance) and I've moved the MoveAgent 
code to this class. This way, it is accessible to other classes, too. 
I'm using it in the BeamEffector, the SoccerRuleAspect, and the 
TrainerCommandParser now.

By the way, the TrainerCommandParser should be ready to be used with 
simspark now, so hopefully, this should make testing the rule stuff a 
little easier :-)

> I don't know if I'm going in the right direction or not! It seems that 
> most of the old code can be used with little or no modification 
> (UpdateKickOff for example). Am I right?!!
>
>

Yes, you're right. Most of the code should work now, so we can 
re-activate the commented out portions of the code little by little and 
test (I already activated the UpdateKickOff part in the SIMSPARK_RULES 
branch of the CVS). One of my main concerns was the performance drop the 
integration of the rules might bring, but I hope my latest commits might 
help to deal with that (crossing my fingers that I didn't break anything 
;-) ).

I'm now working on a new SimControlNode that allows to display GameState 
info in the internal monitor since I'm having a hard time with monitorspark.

If you implement the BoundingBox methods for the agent, maybe you can 
put them in the SoccerBase class. I think they might be useful for other 
classes, too.

Cheers,
Joschka

Re: [simspark-devel] About SoccerRules

[Hedayat V. <hed...@ai...>]

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html;charset=UTF-8" http-equiv="Content-Type">
</head>
<body dir="ltr" bgcolor="#ffffff" text="#000000">
<small><font size="+1"><small>Hi Hesham,<br>
OK, from now, I'll (hopefully!) add new function(s) to calculate the
AABB2 or AABB3 around an agent to be used instead of a single position
in functions such as ClearPlayers functions.<br>
<br>
Just for completeness, I've attached my latest soccerrulesaspect.cpp
file.<br>
<br>
Good luck,<br>
Hedayat<br>
</small></font></small><span><br>
<style type="text/css">blockquote {color: navy !important; background-color: RGB(245,245,245) !important; padding: 0 15 10 15 !important; margin: 15 0 0 0; border-left: #1010ff 2px solid;}  blockquote blockquote {color: maroon !important; background-color: RGB(235,235,235) !important; border-left-color:maroon !important}  blockquote blockquote blockquote {color: green !important; background-color: RGB(225,225,225) !important; border-left-color:teal !important}  blockquote blockquote blockquote blockquote {color: purple !important; background-color: RGB(215,215,215) !important; border-left-color: purple !important}  blockquote blockquote blockquote blockquote blockquote {color: teal !important; background-color: RGB(205,205,205) !important; border-left-color: green !important}</style><i><b>Hesham
<a class="moz-txt-link-rfc2396E" href="mailto:hes...@gm...">&lt;hes...@gm...&gt;</a></b></i> wrote on 06/09/2007 07:50:09
AM:</span><br>
<blockquote   style="border-left: 2px solid rgb(16, 16, 255); color: navy; background-color: rgb(245, 245, 245); padding-left: 15px;"   cite="mid:1d7...@ma..."   type="cite"><br>
Hi Hedayat,<br>
  <br>
Could you please let me know which functions you're working on, so I'll
give the others a try.<br>
  <br>
Cheers,<br>
Hesham<br>
  <br>
  <div><span class="gmail_quote">On 6/8/07, <b class="gmail_sendername">
Hedayat Vatankhah</b> &lt;<a moz-do-not-send="true"   href="mailto:hed...@ai...">hed...@ai...</a>&gt; wrote:</span>
  <blockquote class="gmail_quote"   style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
    <div dir="ltr" bgcolor="#ffffff" text="#000000">
    <p style="margin-bottom: 0cm; margin-top: 0pt;"><small><font   size="+1"><small>Hi,</small></font></small></p>
    <p style="margin-bottom: 0cm; margin-top: 0pt;">I've changed
MoveAgent
for the new agent (thanks BeamEffector!) and fixed ClearPlayers
functions. I don't know if I'm going in the right direction or not! It
seems that most of the old code can be used with little or no
modification (UpdateKickOff for example). Am I right?!!</p>
    <p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
    </p>
    <p style="margin-bottom: 0cm; margin-top: 0pt;">Thanks!</p>
    <p style="margin-bottom: 0cm; margin-top: 0pt;">Hedayat</p>
    <p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
    </p>
    </div>
    <br>
-------------------------------------------------------------------------<br>
This SF.net email is sponsored by DB2 Express<br>
Download DB2 Express C - the FREE version of DB2 express and take<br>
control of your XML. No limits. Just data. Click to get it now.
    <br>
    <a moz-do-not-send="true"   onclick="return top.js.OpenExtLink(window,event,this)"   href="http://sourceforge.net/powerbar/db2/" target="_blank">http://sourceforge.net/powerbar/db2/</a><br>
_______________________________________________<br>
Simspark Generic Physical MAS Simulator
    <br>
simspark-devel mailing list<br>
    <a moz-do-not-send="true"   onclick="return top.js.OpenExtLink(window,event,this)"   href="mailto:sim...@li...">sim...@li...</a><br>
    <a moz-do-not-send="true"   onclick="return top.js.OpenExtLink(window,event,this)"   href="https://lists.sourceforge.net/lists/listinfo/simspark-devel"   target="_blank">https://lists.sourceforge.net/lists/listinfo/simspark-devel</a><br>
    <br>
    <br clear="all">
  </blockquote>
  </div>
  <br>
</blockquote>
</body>
</html>

Re: [simspark-devel] About SoccerRules

[Hesham <hes...@gm...>]

Hi Hedayat,

Could you please let me know which functions you're working on, so I'll give
the others a try.

Cheers,
Hesham

On 6/8/07, Hedayat Vatankhah <hed...@ai...> wrote:
>
>  Hi,
>
> I've changed MoveAgent for the new agent (thanks BeamEffector!) and fixed
> ClearPlayers functions. I don't know if I'm going in the right direction or
> not! It seems that most of the old code can be used with little or no
> modification (UpdateKickOff for example). Am I right?!!
>
>
>  Thanks!
>
> Hedayat
>
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Simspark Generic Physical MAS Simulator
> simspark-devel mailing list
> sim...@li...
> https://lists.sourceforge.net/lists/listinfo/simspark-devel
>
>
>

[simspark-devel] About SoccerRules

[Hedayat V. <hed...@ai...>]

Hi,

I've changed MoveAgent for the new agent (thanks BeamEffector!) and 
fixed ClearPlayers functions. I don't know if I'm going in the right 
direction or not! It seems that most of the old code can be used with 
little or no modification (UpdateKickOff for example). Am I right?!!


Thanks!

Hedayat

Re: [simspark-devel] MonitorSpark Crashes

[Hedayat V. <hed...@ai...>]

Hi Joschka,

/*Joschka Boedecker <jbo...@un...>*/ wrote on 06/04/2007 
06:59:08 PM:
> ...
> I can't reproduce the crash on my Ubuntu 7.04 laptop (Ruby 1.8.5 
> (Ubuntu package), ODE 0.8 (compiled from source) and gcc 4.1.2 (Ubuntu 
> package)). With the latest CVS version, everything runs fine for me. I 
> use standard configure options though, I can try to set debug to no 
> tomorrow.
> ...


I'm using the latest CVS version too. I'm using Ruby 1.8.4 (FC5 package) 
(while the INSTALL file says that it doesn't work, It's worked for me 
till now!), ODE 0.5 and gcc 4.1.0 (FC5 package).

The other systems which I've checked where FC6 with it's standard 
packages and ODE 0.8 I think (not sure about this, they might use ODE 0.5).

Thanks,
Bye

Re: [simspark-devel] MonitorSpark Crashes

[Joschka B. <jbo...@un...>]

Hi Hedayat,

Hedayat Vatankhah wrote:
>
> Hi,
> I can't run monitor spark, it crashes. I've sent the full report as a 
> bug in sserver.sf.net.
>

Thanks for the bug report.

> I've tested it on many computers but it crashes in all of them. I 
> think that all of them were Fedora. Personally I use FC5. Also, I 
> configure my server using --enable-debug=no option.
>
>

I can't reproduce the crash on my Ubuntu 7.04 laptop (Ruby 1.8.5 (Ubuntu 
package), ODE 0.8 (compiled from source) and gcc 4.1.2 (Ubuntu 
package)). With the latest CVS version, everything runs fine for me. I 
use standard configure options though, I can try to set debug to no 
tomorrow.

> Today, I've tried to find the source of the problem. The problem is 
> related to this line in both sparkSetupMonitor & 
> sparkSetupMonitorLogPlayer (in spark.rb):
>
> rubySceneImporter.setUnlinkOnCompleteScenes(true);
>
>
> With this line, the monitorspark crashes on this line in soccersim.rb:
>
> line 57: scene.importScene('rsg/agent/soccer.rsg')
>
>
> The problem arises in RubySceneImporter::ParseScene:
>
> line 195: root->UnlinkChildren();
>
>
> It seems that it causes the dBodyDestroy() function to be called on a 
> previously destroyed body.
>
> When I comment the rubySceneImporter.setUnlinkOnCompleteScenes(true); 
> line in spark.rb, the monitorspark doesn't crash anymore.
>

What version of ruby and ODE do you use? Also, what gcc version of gcc 
do you use? Just to get a better picture...

Cheers,
Joschka

[simspark-devel] MonitorSpark Crashes

[Hedayat V. <hed...@ai...>]

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
</head>
<body dir="ltr" bgcolor="#ffffff" text="#000000">
<p style="margin-bottom: 0cm; margin-top: 0pt;"><small><font size="+1"><small>Hi,<br>
</small></font></small>I can't run monitor spark, it crashes. I've sent
the full report as a bug in sserver.sf.net. I've tested it on many
computers but it crashes in all of them. I think that all of them were
Fedora. Personally I use FC5. Also, I configure my server using
--enable-debug=no option.</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">Today, I've tried to
find the source of the problem. The problem is related to this line in
both sparkSetupMonitor &amp; sparkSetupMonitorLogPlayer (in spark.rb):</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">rubySceneImporter.setUnlinkOnCompleteScenes(true);</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">With this line, the
monitorspark crashes on this line in soccersim.rb:</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">line 57:
scene.importScene('rsg/agent/soccer.rsg')</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">The problem arises in
RubySceneImporter::ParseScene:</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">line 195:
root-&gt;UnlinkChildren();</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">It seems that it causes
the dBodyDestroy() function to be called on a previously destroyed body.</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">When I comment the
rubySceneImporter.setUnlinkOnCompleteScenes(true); line in spark.rb,
the monitorspark doesn't crash anymore.</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">Cheers,</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">Hedayat<br>
</p>
</body>
</html>

Re: [simspark-devel] Countdown for Atlanta

[Joschka B. <jbo...@un...>]

Hi Yuan,

Yuan Xu wrote:
> Hi Joschka,
>
>
>> >>  .... If the
>> >> application is too busy for SDL to create appropriate events, do you
>> >> think it would make sense to have an other application ("referee") 
>> that
>> >> connects to the server as a monitor and sends commands using the 
>> trainer
>> >> protocol? This way we could make the game kick of and do drop ball at
>> >> least...
>> >>
>
> I found this in http://www.libsdl.org/intro.en/whatplatforms.html
>
> "Linux and BeOS support the SDL_INIT_EVENTTHREAD flag which, when
> passed to SDL_Init(), asks the event loop to run asynchronously in
> another thread. This is useful for color cursors that respond even
> when the application is busy."
>
> But initializing SDL with SDL_INIT_EVENTTHREAD can not get clear 
> effort ;-(

It sounded very helpful, but my tests so far also didn't get a clear 
result. In fact, the input can be best controlled in single-thread mode 
without using the SDL_INIT_EVENTTHREAD mode for me so far.

By the way, I think I'll write an email to the list today, asking for 
feedback on the timer. It's important to hear the teams' experience 
before we make a decision for Atlanta, I think.

>
> If the monitor and server run on one machine, it seems OK.
> But I have some problems while connecting the monitor via the net.
> If there is no agent, the monitor works OK.
> If one agent connected, and the full scene message is received, but
> the prefix size of message can not read correctly!
> It reminds me what happened in the agentspark before.
>
> Is there anyone know the solution?
>

Yes, this definitely sounds like a similar (or the same problem), but I 
didn't have time to check yet. Can anybody else observe the same problem?

Cheers,
Joschka

Re: [simspark-devel] log player - report 2

[Joschka B. <jbo...@un...>]

Hi Hesham,

Hesham wrote:
>
> Instead of defining a variable in one of the script files, in spark.rb 
> after creating an object of SparkMonitorLogFileServer, a function (of 
> the object) is called to set the length of the delay between steps.
>

Thanks for the code.

> Hesham
>
> On 6/2/07, *Hesham* <hes...@gm... 
> <mailto:hes...@gm...>> wrote:
>
>
>     Hi Joschka
>
>     I've attached the changed files for the log player (based on CVS
>     version May 27). I'm a bit confused about the CVS branches and
>     didn't know which one I should use to make this patch, so I send
>     the files.
>

I committed it to the main branch of the repository yesterday.

>     In this version:
>        p : pauses the log player (toggle)
>        f : goes to next step
>        b : goes to previous step
>        l : toggles between forward playback and backward playback
>

Works nicely :-)

Cheers,
Joschka

[simspark-devel] ForceResistancePerceptor Finished (hopefully!)

[Hedayat V. <hed...@ai...>]

Hi,

I've finished my work on ForceResistancePerceptor. Also, I've added it's 
documentation to TEXT_INSTEAD_OF_A_MANUAL.txt (please read it as my 
English is not good  and it might need editing ;)

It need some testing, but hopefully it should work well! (For using this 
perceptor, TouchPerceptorHandler should be used instead of 
ContactJointHandler).

The final output is an weighted average position (in local coordinates 
relative to the body) and the total force vector applied to it 
(previously I've suggested average force vector, but it seems to be 
completely wrong!)


A small question: I've used GetLocalTransform().Pos() from the Transform 
class to convert global coordinates to local coordinates. But it seems 
that using GetWorldTransform() is more appropriate (while the result is 
the same). But what about the body object below the transform node? 
Should I use the Body::GetPosition() method of that function instead? 
Which choice is the best? (and why? (optional!:))


I should be prepared for both my university exams; but I'll try to work 
on soccer rules as much as possible. I'll switch to the SIMSPARK_RULES 
branch now!


Good luck,

Bye

Hedayat

Re: [simspark-devel] log player - report 2

[Joschka B. <jos...@am...>]

Hi Hesham,

Hesham wrote:
>
> Instead of defining a variable in one of the script files, in spark.rb 
> after creating an object of SparkMonitorLogFileServer, a function (of 
> the object) is called to set the length of the delay between steps.
>

Thanks for the code.

> Hesham
>
> On 6/2/07, *Hesham* <hes...@gm... 
> <mailto:hes...@gm...>> wrote:
>
>
>     Hi Joschka
>
>     I've attached the changed files for the log player (based on CVS
>     version May 27). I'm a bit confused about the CVS branches and
>     didn't know which one I should use to make this patch, so I send
>     the files.
>

I committed it to the main branch of the repository yesterday.

>     In this version:
>        p : pauses the log player (toggle)
>        f : goes to next step
>        b : goes to previous step
>        l : toggles between forward playback and backward playback
>

Works nicely :-)

Cheers,
Joschka

Re: [simspark-devel] Countdown for Atlanta

[Yuan X. <xuy...@gm...>]

Hi Joschka,


> >>  .... If the
> >> application is too busy for SDL to create appropriate events, do you
> >> think it would make sense to have an other application ("referee") that
> >> connects to the server as a monitor and sends commands using the trainer
> >> protocol? This way we could make the game kick of and do drop ball at
> >> least...
> >>

I found this in http://www.libsdl.org/intro.en/whatplatforms.html

"Linux and BeOS support the SDL_INIT_EVENTTHREAD flag which, when
passed to SDL_Init(), asks the event loop to run asynchronously in
another thread. This is useful for color cursors that respond even
when the application is busy."

But initializing SDL with SDL_INIT_EVENTTHREAD can not get clear effort ;-(

If the monitor and server run on one machine, it seems OK.
But I have some problems while connecting the monitor via the net.
If there is no agent, the monitor works OK.
If one agent connected, and the full scene message is received, but
the prefix size of message can not read correctly!
It reminds me what happened in the agentspark before.

Is there anyone know the solution?


-- 
Best wishes!

Xu Yuan
School of Automation
Southeast University, Nanjing, China

mail:   xuy...@gm...
          xy...@ya...
web:   http://xuyuan.cn.googlepages.com
--------------------------------------------------

Re: [simspark-devel] New touch sensor

[Hedayat V. <hed...@ai...>]

Hi,


/*Joschka Boedecker <jbo...@un...>*/ wrote on ۰۷/۰۵/۳۱ 08:12:28:
>>> Since we are trying to decrease output to the agents to a minimum 
>>> currently, I would propose to output the average of the force vector 
>>> magnitudes. What do you think?
>> OK. I'm agree. What about the (weighted) average position? I think 
>> it'll be really helpful!
>>
>
>
> Sounds good. I also think it can be very useful. Could you please also 
> write a short documentation for this perceptor (in the style of the 
> documentation in the TEXT_INSTEAD_OF_A_MANUAL file in the doc 
> directory)? This would be very helpful.
OK. I'll do it when I finished working on this sensor.

Bye,
Hedayat

>
>
> Thanks,
> Joschka