It makes a heavy use of messages to exchange information about the battle field: friends positions, enemy positions, bullets fired, enemies targeted, and hits by enemy fire (I used it to determine which mate was an enemy targeting).
It got an average classification in the F2F Teams competition.
It uses a kind of antigravity system. It feaures also some functionalities especific for teams:
It has an hybrid Virtual Bullets/pattern matcher gun. The gun owes a lot to Paul Evan's SandboxMini, from where I got inspiration. The most remarcable characteristics are its hability to combine VB and pattern matching, and its hability to interpolate missing data points (using a linear function). Also, the gun receives information from all the team mates, so it can learn for every bullet fired by the team (not only the bot).
The bad news is that the results never satisfiied me completely.
No dodging at all, except for friendly fire.
It has only a team strategy. It tries to combine the fire to the densiest enemy concentrations. It uses a weigthing fnction to decide if it will fire to the team target or to a direct threat (ie. a very close enemy).
To avoid getting stationary when only one enemy is left, I add some random moving antigrav-points.
It's just the name of another poisonous snake, as Aspid.
I'm not releasing the code, but I have no problem to share the ideas, so just ask.
I don't think I'w improve it. There is a non released version that may be I'w reease if I decide it does better that the current one. But (probably because it was my first attempt) the code is quite messy and I preffer to develop a new team using the lessons learn from MambaTeam.
NA.