Multi-Agent Surveillance

A more descriptive readme file will follow in phase 3 :)

Gamemodes

Exploration

The Guards have to explore as much of the map as they can. The game ends if no coverage progress is made in a certain amount of time.

Guards vs. Intruders

The game ends when either the guards have captured all intruders, or when all alive intruders have been to the target area at least once.

Agent types

• Random
  Constantly walking while changing rotation arbitrarily.
• Remote
  Controllable by the user. Move forward using W and turn left or right with A and D.
• Bug-0
  Exploration agent, has a specific target direction in mind and tries to walk to it.
• Ant colony
  to be elaborated upon in phase 3...
• NEAT
  Movement & rotation determined via a neural network. This is a pretty neat agent...

Running

With IntelliJ, a GUI instance can be started via the 'run' task in the Gradle menu. If not using IntelliJ, simply use the command gradle run in the main directory.

Suggested setup:

Item	Value
Gamemode	Guards vs Intruders
Guard type	NEAT Agent
Intruder type	Bug Agent
Map	File map (phase2_2.txt)

Related stuff

• M.A.S. Map Editor, used to create custom maps in our experiments.

Authors

Martin Aviles, Lena Feiler, Willem Ploegstra, Konstantin Sandfort, Christopher Schiffmann