Adaptive Hierarchical Model Predictive Control via Dynamic Bayesian Learning for Autonomous Drone Swarms in Dynamic Environments

**Abstract:** This paper introduces a novel Adaptive Hierarchical Model Predictive Control (AH-MPC) framework for autonomous drone swarm coordination in dynamic and unpredictable environments. By integrating Dynamic Bayesian Learning (DBL) with hierarchical MPC, we enable robust and adaptive trajectory planning and execution for a multi-agent system facing changing environmental constraints and inter-agent dynamics. Our approach leverages real-time sensory data to continuously refine predictive models, facilitating efficient swarm navigation, obstacle avoidance, and task execution while accounting for both internal and external uncertainties. Unlike traditional MPC methods, AH-MPC demonstrably enhances robustness to disturbances and reduces computational complexity by distributing control tasks across a hierarchical structure. This paradigm shift promises substantial improvements in autonomous drone swarm applications ranging from search and rescue to infrastructure inspection.