Human-aware Motion Planning for Aerial Robots

Abstract: My thesis focus on the challenge of autonomous UAV navigation in dynamic environments, particularly when interacting with moving humans. By combining Social Generative Adversarial Networks (SGAN) for human trajectory prediction and the Rapidly-exploring Random Tree Star (RRT*) algorithm for real-time path planning, this project enhances the UAV's ability to anticipate and avoid potential collisions. The SGAN model predicts human movement based on historical trajectories, while RRT* continuously replans the UAV's path to ensure safe navigation in unpredictable settings.

The approach developed in this project has significant practical applications in areas where human-robot interaction is crucial.  By integrating real-time human trajectory prediction with adaptive path planning, this method improves both the safety and efficiency of UAV operations in complex, real-world environments.