Evolution of Agent Coordination in an Asynchronous Version of the Predator-Prey Pursuit Game

Adapting the Predator-Prey Game Theoretic Environment to Army Tactical Edge Scenarios with Computational Multiagent Systems

“…As an intermediate step, some authors have extended (while maintaining) the discrete environment with diagonal movements [32], or more complex cell-shapes (such as hexagons [18,32], or irregular convex cells [33]). The timing of predator and prey movement has also been manipulated, with the standard parametrization allowing agents to move synchronously per time step [32], alternating agents' movements through a sequence of time steps (e.g., turn-taking as described in [34]), or allowing more realistic unrestricted, asynchronous movement [23] such that predators and prey can react to each other at flexible time intervals.…”

“…The original predator-prey pursuit environment ( Figure 1) required four predator agents to surround the single prey agent from four directions in a discretized grid world [1,9,21,23]. The predator agents were guided by an algorithm and their movements were limited to one grid square per time step to an adjacent available square (not occupied by another agent or a boundary) in only the vertical or horizontal directions (no diagonal movements).…”

“…Cooperative algorithms inherently enable predators to collaborate [12,[24][25][26][27][28]. However, under certain conditions, predators using a greedy strategy may have greater success [29], though cooperative algorithms often win out when a sophisticated prey is faced [23], and in some environments prey benefit most from a mix of greedy and altruistic strategies [20]. These algorithmic explorations of multi-agent cooperation show how the predator-prey pursuit environment provides an ideal testbed for understanding collaborative agent behavior.…”