Effektiviteten hos kommunikationsmodeller för jägare i rutnätsbaserade jakt-flyktspel

Pursuit-evasion games (PEGs) are played between two teams with opposite goals: the pursuers should capture the evaders, and the evaders should avoid being captured. There are many implementations of PEGs with different rules and map layouts. In this thesis, PEGs are simulated on a two-dimensional grid containing randomly placed obstacles. The players start in random positions, with pursuers having limited vision and being able to communicate both their observations and deduced knowledge with other pursuers. We propose three communication models – a corridor-based, a radius-based, and a time-interval-based model – and test their effectiveness by measuring the time to capture all evaders. To determine which model is preferred, they were evaluated against each other on open and dense grids by varying the number of obstacles. We found that all three models perform similarly, with the time-interval-based model performing slightly worse when the game consists of many players. Furthermore, we found that it is important for pursuers to communicate frequently. However, these conclusions apply to the specific PEG studied, which means the communication models may produce other results in other game setups.