Vrednotenje odpornosti in učinkovitosti agregiranega programiranja v rojih robotov

This master's thesis presents the design, implementation, and evaluation of a decentralized aggregate programming algorithm for dynamic resource management in robot swarms. Leveraging the Field Calculus++ (FCPP) framework, the algorithm facilitates self-organizing and resilient swarm behaviors that optimize resource allocation for applications in military operations, agriculture, search and rescue, and industrial inspection. The core contribution is an algorithm enabling 'worker' drones to dynamically exchange supporting 'scout' drones based on evolving task requirements. This fully decentralized approach ensures optimal resource distribution without a central coordinator. The system architecture integrates a custom Aggregate Programming (AP) Engine with ROS 2 middleware and the Crazyflie drone platform. Validation was performed in both the Gazebo simulation environment and on physical hardware. Experimental results demonstrate the algorithm's effectiveness across diverse scenarios, including varied trajectories, drone quantities, and failure conditions. The system successfully maintained circular formations and dynamically reassigned scouts to balance 'scout need' across the swarm. Although challenges such as propeller downwash instability were observed in physical tests, the findings confirm the viability and resilience of aggregate programming for complex, real-time swarm coordination. Future work will focus on direct on-board deployment of the algorithm and the integration of more advanced collision avoidance techniques.