DEAMN: Advanced Multi-Drone Navigation and Coordination for Dynamic Environments Using ROS-Based System

This paper describes a complete approach to the problem of navigation for the multi-drone system in complex environmental conditions using the functional programming language Robot Operating System (ROS). The approach we propose, called the Dynamic Environment Adaptive Multi-Drone Navigation (DEAMN), integrates sensors, which are very high technology, environmental mapping based on real-time data, and control strategies that seem to be adaptive to a degree to obtain such a result as precision localization, environment awareness and without any friction from the sim-world to the real-world transition. The main aspects are sensor fusion & localization, dynamic environment mapping, multi-agent coordination, realtime path planning, and adaptive feedback control. The DEAMN framework employs the most recent technologies, such as Extended Kalman Filters, Occupancy Grid Mapping, and Model Predictive Control, to guarantee the best possible performance in such a complex architecture and rapidly changing conditions. We also focus on the necessity of the effective cooperation of multiagents by using a consensus algorithm and the smooth transition from the simulation to the real-world system through domain randomization and model transfer methods. The experimental results of our research highlight considerable enhancements in localization accuracy, mapping resolution, and collision avoidance success rates, as well as reduced path planning time and Sim-toReal deviation. The DEAMN algorithm shows high adaptability and reliability in real-world scenarios, therefore, it is undoubtedly a competent solution for advanced multi-drone operations in dynamic environments.