Gliders2012: Tactics with Action-dependent Evaluation Functions

– distributed client/server system running on a network, leading to fragmented, localized and imprecise (noisy and latent) information about the environment (field) [2]; – concurrent communication with a medium-sized number of agents [3]; – heterogeneous sensory data (visual, auditory, kinetic) and limited range of basic commands/effectors (turn, kick, dash, . . .) [4]; – asynchronous perception-action activity and limited window of opportunity to perform an action [5]; – autonomous decision-making under constraints enforced by teamwork (collaboration) and opponent (competition) [6]; – conflicts between reactivity and deliberation [7]; – no centralized controllers and centralized world model (no global vision, etc.) [8, 9].