Multi-agent learning approach to dynamic security patrol routing

Patrols are groups of security personnel, such as police officers or soldiers, whose main job is patrolling an area to maintain peace. In this study, we simulate their activities in an artificial urban environment similar to a real city that has banks, shops, and other hotspots that may attract crime. It is believed that specific patrol routes have influence in reducing crime rates. We propose a multi-agent-based XCS learning classifier system implementation to generate their behaviors to learn better route to prevent a possible crime outbreak in the neighborhood.