Zdecentralizowany system zarządzania zadaniami dla zespołu robotów w misji poszukiwawczo-ratowniczej

Saving human lives is an important and noble application of robotics. Managing the actions of distributed robots is essential for the coordinated execution of a rescue operation. The goal of this work is to develop a solution for task management within search and rescue mission and to evaluate the use of a decentralized approach, which enhances the system’s resilience and scalability. The system was developed to coordinate robot actions in a designated area search task. Each distributed component of the system is assigned to a single robot and integrates with its control software. The absence of a single, central module increases the system’s resistance to complete failure. A task delegation algorithm was designed, defining the communication scheme and consensus protocol used by the distributed components. Tasks are assigned in a way that minimizes the overall mission completion time. Additionally, the system allows for defining dependencies between assigned tasks, which can be used to maintain robot formations. To address temporary communication failures, an algorithm for knowledge synchronization was developed. This algorithm improves the system’s robustness and liveness when deployed in environments where stable network connections between robots cannot be guaranteed. The implemented system was tested in the Gazebo simulator within a scenario involving area exploration. A qualitative analysis based on test scenarios, as well as a quantitative analysis focused on scalability, was presented. It was observed that decentralized systems are highly dependent on communication mechanisms, making reliable communication essential for correct operation. Moreover, it was found that the process responsible for generating tasks has significant influence on system performance.