Navigation in Non-Static Environments with Autonomous Drones: A Kalman Filter Reinforcement Learning Approach

Autonomous drone systems have grown in various industries, but their effectiveness in dynamic environments remains challenging. This study addresses the issues faced in path planning and state estimation for drones operating in non-static environments. To tackle these challenges, a solution combining Kalman Filters and Advanced Reinforcement Learning (RL) Algorithms is proposed. Three RL algorithms are compared to evaluate their performance. Combining the Kalman Filter and RL techniques improves path planning and decision-making, resulting in successful navigation in simulated dynamic scenarios. The approach is designed for continuous state-action environments.