[Movement] Going around a friendly agent

[rafaelcaue]

At the moment the most simple solution I could think of was each agent picks a random location between the original position it wanted and the other remaining two (after removing the opposite direction to the original).

This is of course not efficient or smart, however it is not trivial to know which agent it is even if I have already identified it. I see two possible solutions:

• call identify again to identify the agent based on local view (or global, depends which one would cost less)
• keep all agents positions on the team artifact (this is tricky to keep synchronised...)

I don't think I like either one of those...

[rafaelcaue]

Because of the increased number of obstacles, it is now possible to get two agents stuck trying to get past one another.