|
From: Mahdi <zig...@gm...> - 2009-03-03 12:57:07
|
Hi All C B wrote: > Hello Feng, > No, it happens when connecting the agent, after beaming... > at least in my case, and it happens just sometimes. > After connecting several times, suddenly one of those connections > gives the jump behavior, the agent gets fired to the sky and > then falls very far. > > Has this ancient problem been solved? I have read about no messages that say this jump behaviour is solved. The problem is occuring for me a lot of times, whenever the agent sends the beam command. Following the jump is the simulators skipping remaining time messages, and Agent gets illegal messages, like the one I provided before with a lot of closing brackets at the end of message with no corresponding opening brackets. > Sander thinks that it could be that the agent is beaming a little > under the floor. > > If this is the case, and since agents do not provide Z coordinate for beaming, I suggest that the server side beam command actually beams agent a little higher, like 2 or 3 centimeters above the ground. I suspect that the problem is another thing: The simulator recvs the beam command, but actual beaming occurs in the next cycle of simualator's loop, following that simulators sent messages collide between cycles, and the problems occur. I barely suspect this may be the problem, since the simulators main loop is clear. Is simulator currently multithreaded? Another thing which is still an issue, is that flags are really not enough for agent localization, when agent is looking towards any goal, any flag that he sees are in a line, which makes localization very hard and consumes a lot of process time. Since agent does not have vision of the field, more flags should be added to the field borders. The part in see messages which is responsible to inform the agent of his body parts locations, does not have a unique identifier, I had a little talk with hedayat, and we came to the suggestion to add an identifier to the message like 'bd', so that in parsing this part of message would be recognized better and safer. Consider adding more data parts to the vision message may collide with the body part vision message. Cheers Mahdi > Best regards, > Carlos Bustamante > Borregos3D > > > >> De:: Feng Xue <hen...@ma...> >> Asunto: Re: Re: [Sserver-three-d] Agent Jump bug >> A: "kam...@ya..." <kam...@ya...>, "athom" < at...@12...> >> Cc: "Sse...@li..." < Sse...@li...> >> Fecha: miércoles, 11 junio, 2008, 8:18 pm >> Hi athom and C B, >> >> Did it happen when the agent collides the ball with >> enabled inner collision? Or would you please summarize >> this problem to help us fix the bug? >> >> Thank you in advance! >> >> >> >> >> Best Regards, >> Feng Xue >> 2008-06-12 >> >> >> >> Sent By: C B >> Sent At: 2008-06-12 04:49:02 >> Sent To: sse...@li... >> CC To: >> Topic: Re: [Sserver-three-d] Agent Jump bug >> >> >> Hello, >> I have seen this behavior too, a lot of times, with the new >> server and the Nao model. >> What could it be? Seems to be non-deterministic. >> >> Best regards, >> Carlos Bustamante >> Borregos3D >> >> De:: yeer <at...@12...> >> Asunto: [Sserver-three-d] Agent Jump bug >> A: "sse...@li..." >> <sse...@li...> >> Fecha: martes, 10 junio, 2008, 8:22 am >> >> >> hi,guys >> It seems the agent sometimes will jump high into the sky >> when I run the game with one agent. >> at the same time it prints errors as follows: >> ODE Message 3: LCP internal error, s <= 0 >> (s=0.0000e+00) >> >> >> >> >> >> >> Yahoo! Deportes Beta >> ¡No te pierdas lo último sobre el torneo clausura 2008! >> Entérate aquí http://deportes.yahoo.com >> > > > ____________________________________________________________________________________ > Yahoo! Deportes Beta > ¡No te pierdas lo último sobre el torneo clausura 2008! Entérate aquí http://deportes.yahoo.com > > ------------------------------------------------------------------------- > Check out the new SourceForge.net Marketplace. > It's the best place to buy or sell services for > just about anything Open Source. > http://sourceforge.net/services/buy/index.php > _______________________________________________ > Sserver-three-d mailing list > Sse...@li... > https://lists.sourceforge.net/lists/listinfo/sserver-three-d > |