Autonomous pursuit evasion system for mini Unmanned Aerial Vehicles

As a result of the widespread use of unmanned aerial vehicles, UAVs are actively used in many different missions in both civilian and military fields. It is anticipated that UAVs, which are primarily used for reconnaissance and surveillance, homeland security, and close air support at the tactical and strategic level in the military field, can also be used in air-to-air combats in the near future. Therefore, UAVs with technological equipment that can engage with other UAVs will be in an advantageous position. The ability to perform air-to-air combats autonomously will provide advantages in terms of both performance and time-criticality. In terms of Mini or Small UAVs, air-to-air combats are mainly focused on evasion and pursuit missions. There are several pieces of research about path planning missions with avoidance zones in literature. There are also several research about pursuit-evasion problems. However, there is no research that combines the two cases to the best of our knowledge. In this thesis, the problem of path planning in avoidance zones and the problem of pursuit-evasion missions are merged, and a solution is proposed for the integrated problem. Mission area is divided into green, gray, and red areas according to the level of danger. In this thesis, a mini unmanned aerial vehicle software that can autonomously perform escape and pursuit missions in different types of zones has been developed. Mission Control Software uses Reinforcement Learning and Q Learning to perform autonomous missions.