Optimizing Unmanned Aerial Vehicle Path finding in Complex Urban Airspaces: A Hybrid Genetic Algorithm with Bidirectional and Greedy Initialization

Abstract

Energy-efficient solutions are crucial for sustainable urban logistics, especially with the increasing reliance on Unmanned Aerial Vehicles (UAVs) in congested urban environments. Effective path planning in such scenarios must address challenges like high traffic density, obstructions, and energy constraints while ensuring safety and efficiency. This study focuses on aerial path planning for autonomous air mobility in obstacle dense low altitude urban air space, integrating energy-efficient strategies with optimal path planning techniques. Genetic Algorithms (GAs) are powerful optimization techniques known for their ability to efficiently explore large and complex solution spaces. Their iterative approach mimics natural selection, making them highly effective in identifying near-optimal solutions for complex, non-linear problems. A modified GA is proposed, incorporating greedy initialization based on Euclidean heuristics to optimize the initial population. Additionally, the initialization process includes a bidirectional approach, assigning half of the population from the goal to the start. These enhancements aim to improve convergence speed, solution quality, and computational efficiency. The performance of the modified GA is evaluated against a standard GA using a random maze generation algorithm designed to simulate urban airspaces with a high obstacle density. Comparative results demonstrate the superiority of the modified GA in navigating complex environments with reduced computational expense and faster convergence. This approach highlights the potential of integrating innovative initialization strategies within GAs to address the challenges of path planning in urban UAV logistics, paving the way for more sustainable and efficient aerial mobility solutions.