Operation oriented swarm software for hybrid robotic systems

Swarm robotics is a dynamic system inspired by nature and composed of robots working together. Studies in this field focus on understanding how decision-making processes and communication can be carried out effectively, along with examining collective behavior and coordination. The development of existing studies in the field of swarm robotics for specific application areas prevents the formation of a generalizable algorithm for multiple operational tasks. This situation negatively affects the reuse of algorithms developed for special tasks in new tasks. In this study, a generalizable algorithm that will ensure the performance of swarm robotic systems in operational tasks and can be used in new tasks with minimal changes has been studied. In the study, it was aimed to reach certain operational target swarms by making an efficient and effective distribution of robots on certain targets whose locations are known. Since the study was working on an algorithm focused on operational distribution, battery constraints and collision situations of the robots were not addressed. In the first stage of the study, a mapping algorithm was proposed that involves regional mapping of targets in a certain operational area, taking into account road layouts and costs. In the second stage of the study, the regional density method used to assign robots to the mapped regions was proposed. In the third stage of the study, a local regional navigation algorithm was proposed that models the navigation of single or multiple robots within regions using the relationship score and target matrix assigned to the targets. In the last stage of the study, a unique topological swarm architecture was proposed, which will enable the original algorithms created to work under a certain topology and form the infrastructure of the swarm software. During the evaluation phase of the study, a 3-stage simulation was made. In the first stage of the evaluation simulations, a simulation study was conducted that included the regional mapping algorithm zoning the target regions with an appropriate distribution and relationship status and was modeled with the Python programming language. In the second stage of the simulations, a simulation study was conducted that included the performances of single and multiple robot assignments on targets within the regions mapped by the local regional navigation algorithm and was modeled with the Python programming language. In the third stage of the simulations, the operation of the algorithm in a simulation environment close to the real environment was tested by using Gazebo and ROS software.