From: Hedayat V. <hed...@gm...> - 2012-12-26 19:17:49
|
Hi Stefan, Sorry for the delay! /*Stefan Glaser <gla...@gm...>*/ wrote on Mon, 17 Dec 2012 13:14:18 +0100: > Hey Hedayat, > > nice to hear from you again! Thanks :), you too > I also planed to contribute a bit to the community this season, but > I'm not sure yet, what I will be able to do. > I certainly want to put some effort to the wiki and pdf manual again. > I just implemented the monitor protocol on my own and recognized some > differences to the wiki, which I'll document soon. Furthermore I also > want to put some new section to the wiki introducing to terms like > "How to use the simulator for machine learning" and also nail down the > configuration files of the simulatior, add the new percrptors and > effectors for stiffness control, etc. > So if you don't have anyone taking care of some documentation, I may > be the one. Also feel free to give me a hint where documentation is > still missing. In general, I think the wiki is not too bad any more. > We never removed finished points of the todo-list, maybe I should > spend some time on this as well ;) That's great. Event if somebody else is also willing to help, it would be great to have you there too. :) I've already added your name to 2013 MC page. > > The next point may be the monitor protocol. As I just sad, I > implemented it and I'm currently also working on an application, > described in the next paragraph. While doing so, I recognized that it > is not very network efficient. Given that the logfile of an 5-6 minute > game is around 120MB with 5 frames per second, I calculated something > around 0.4MB/sec, which is 3.2MBit/s. Given that we have a stream to > the monitor of 25 frames per second, this results in 16MBit/s network > load alone for the monitor protocol. Is this correct? Sounds quite a > lot for me... Given that I at some point want to provide a better > media support in the league and want to use the full 50 frames per > second to create some slow motion video scenes, this would result in > 32MBit/s load with the current protocol. And even worse, if the > application I'm currently programming will be used in future > competitions, we always have 2 monitors connected to one server. > We will need to discuss this point again once I have time to take care > of it and my ideas are good enough to implement them. Or do you > already have an idea how to improve this? I cannot comment on the numbers. But the load increases with the number of players. However, if your application is ready, the monitors can connect to your application rather than the main server. I had some ideas, but I should check the latest in robotics software world to see if they are outdated. So, I agree that we should discuss it soon. > > The program mentioned above I'm currently implementing is a streaming > server and client. I have a lot of friends at home which are > interested at the competitions, but never have time to join them at > place. Or e.g. you this year had no chance to watch the games live. So > I decided to programm a streaming server/client. The idea is quite > simple. The first version of the streaming-server will connect to a > simpark server as monitor and so to say just forward the messages to > the connected streaming clients in bundles. The streaming client > connects to the streaming server and outputs the server messages on a > local port in real time. This way one can connect any common monitor > to the local streaming client and opperate the camera, etc. control > commands are of course not forwarded. This way we are able to build a > stream server hierarchie and stream to interested universities. > Furthermore we can provide further local viewing stations on the > campionchips. Last year for example at the Dutch Open there was a > second monitor connected because the public area was several meters > away from the real competition. I think I will finish and publish the > streaming server and client the end of this year. This may also be a > good time to discuss about the monitor protocol in order to reduce > network load. That's great, specially if it works for slow or unreliable connections too! > > The last point in my list is a training mode of the server in which an > agent is able to beam itself and the ball freely, perceives ground > truth data, etc. Currently a lot of this options are already somehow > available. But the big point is that if we realy want to open up the > simulator for machine learing, we should allow theese options with one > parameter, maybe even a command line parameter. Yes, this is certainly an area in need of improvements and more user friendliness. > > I will write you another email when I have some time to spend on the > server. I will definetly need some help to get startet. Great. Send it to simspark-devel so that others will see it too. They can also provide valuable input. Regards, Hedayat > > Cheers, > Stefan > > > > > > Am 16.12.2012 21:37, schrieb Hedayat Vatankhah: >> Hi all! >> Like previous years, we are going to create a roadmap for the >> development of our soccer simulator till RoboCup 2013 competitions >> specially since TC decisions are already announced. I've create a >> stub at [1], which is going to be used to coordinate our 2013 >> development efforts. Also notice that the development of the >> simulator is NOT restricted to TC decisions and everybody can >> implement new ideas and contribute to the simulator. >> >> Therefore, I invite everyone to join us in MC and contribute in the >> development of the simulator. You can pick up any of the tasks, or >> even propose new ones and work on them. Please contact me if you are >> interested in helping us in the development of the simulator. >> >> However, if you cannot help us in the development or documentation of >> the simulator, you still can help. There are a number of open issues >> in TC proposal, which should be addressed as early as possible so >> that we will have a usable design and implementation. Please read the >> proposal carefully and send your feedback. >> >> Regards, >> Hedayat >> >> [1] >> http://simspark.sourceforge.net/wiki/index.php/Maintenance_Committee_2013 >> >> >> >> ------------------------------------------------------------------------------ >> LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial >> Remotely access PCs and mobile devices and provide instant support >> Improve your efficiency, and focus on delivering more value-add services >> Discover what IT Professionals Know. Rescue delivers >> http://p.sf.net/sfu/logmein_12329d2d >> >> >> _______________________________________________ >> Simspark Generic Physical MAS Simulator >> simspark-devel mailing list >> sim...@li... >> https://lists.sourceforge.net/lists/listinfo/simspark-devel > |