SynnalagmaQuestion

Controling Robocode

I'm trying to write a controller for robocode but I always have this error message. Does anybody know what can cause this ? I use approximatively the same code as RoboLeague to have a blocking robocode engine. The error only occur after something like 70 rounds (7 battles) (not exactly the same number each time) and battles are one-on-one with sample.Tracker sample.MyFirstRobot and sample.Target. If you have any idea please post a response here.

Sometime it's this error message :

Here is the error message :
****************
Another exception has been detected while we were handling last error.
Dumping information about last error:
ERROR REPORT FILE = (N/A)
PC                = 0x6D3CF788
SIGNAL            = -1073741819
FUNCTION NAME     = JVM_FindSignal
OFFSET            = 0x186F8
LIBRARY NAME      = C:\Program Files\Java\jre\bin\client\jvm.dll
Please check ERROR REPORT FILE for further information, if there is any.
Good bye.
Unexpected Signal : EXCEPTION_ACCESS_VIOLATION occurred at PC=0x6D3CF788
Function=JVM_FindSignal+0x186F8
Library=C:\Program Files\Java\jre\bin\client\jvm.dll
Current Java thread:
        at robocode.peer.robot.EventManager.onCustomEvent(EventManager.java:552)
        at robocode.peer.robot.EventManager.processEvents(EventManager.java:758)

or sometime it's this one :

        
****************
Another exception has been detected while we were handling last error.
Dumping information about last error:
ERROR REPORT FILE = (N/A)
PC                = 0x6D3CF788
SIGNAL            = -1073741819
FUNCTION NAME     = JVM_FindSignal
OFFSET            = 0x186F8
LIBRARY NAME      = C:\Program Files\Java\jre\bin\client\jvm.dll
Please check ERROR REPORT FILE for further information, if there is any.
Good bye.
Unexpected Signal : EXCEPTION_ACCESS_VIOLATION occurred at PC=0x6D3CF788
Function=JVM_FindSignal+0x186F8
Library=C:\Program Files\Java\jre\bin\client\jvm.dll
Current Java thread:
        at java.lang.Throwable.fillInStackTrace(Native Method)
        at java.lang.Throwable.<init>(Throwable.java:180)
        at java.lang.Error.<init>(Error.java:37)
        at robocode.exception.EventInterruptedException.<init>(EventInterruptedException.java:10)
        at robocode.peer.robot.EventManager.processEvents(EventManager.java:699)

Please post any response :

You could use the battles engine for RR@H. It is already done and very simple :-) -- Albert

I have the same problem with RR@H. I guess it's a problem with unsynchronized thing, but I don't know wich one... -- Synnalagma

Representatives Bots

I'm looking for some typical bot in the following domains :

 - mouvement (a bot representative of a categorie like RandomMovement)
 - targetting 

I need them in order to achieve some training for NeuralReinforcementLearningMovement.

please enter some names here --Synnalagma

How about Frankie for movement: is features random movement and has a high score in the MovementChallenge (see MovementChallenge/MCIndex) But what do you mean with a 'typical bot in the targetting domain'? -- Loki

I mean a bot that can be representative of a certain type of targetting like a bot using only circular targetting or only guess factor targetting. The problem is that i need to collect data to train my movement. the trainer is an external program taking like 30 hours to run (so you need to be sure about what you're doing). --Synnalagma

Frankie uses PatternMatching only for its targeting. In fact it's a bot completely created from the requirements of the PatternMatchingChallenge? and the MovementChallenge so it should be a very typical bot in both categories. Another typical bot would be Gouldingi which uses a different RandomMovement from Frankie and also serves as a proof of concept bot for AngularTargeting/Factored. And what about SandboxDT which has the best movement in the population and also uses uncompromised StatisticalTargeting? -- PEZ

The problem is that I need representative not the best one (since the risk is that it can do it good with SandboxDT and very poor with all other bot). Thank for your responses. --Synnalagma

I doubt that. DT is the least predictable bot around. If you can sort it out you can sort any bot out. But go with Frankie then, it's not the best anywhere, yet representative of RandomMovement and PatternMatching aim. -- PEZ

No, I agree with him that one could make a bot that does good with SandboxDT and poor with everyone else. Not because of weaknesses in his movement so much as ways that his aim could be semi-'fooled'. I was telling nano the other day how I made a bot that manipulates its enemy's aim and fires back with head-on aim only when he sees his enemy fire (and at the same power), and this robot did about as well against DT as Cigaret does for the first 50 or so rounds. 290 bytes. Not a worthwhile bot against anyone normal, though. -- Kawigi

I'm not sure I quite follow. But I still think that if you can make a bot that defeats DT you will also defeat most other bots. Besides, from what I have seen on the latest seasons of EternalRumble and my own tests, Marshmallow is the bot giving DT most trouble at the moment. Marshmallow is also a #2 contender, and so is of course Cigaret which also gives DT a fight. As long as you're using sensible techniques you can hardly go wrong concentrating on defeating DT. -- PEZ