Real-Time Trajectory Planning Framework for UAV in Unknown and Dynamic Environment
Zheyuan Mei · 2024
This paper introduces a novel trajectory planning framework for unmanned aerial vehicles (UAV) navigating in unknown and dynamic environments. Our approach uses an enhanced Simple Online and Realtime Tracking (SORT) algorithm adapted for three-dimensional space to track dynamic obstacles in real time. Future trajectories of these obstacles are predicted by Bezier curves, with an analytical solution obtained through the Lagrange multiplier method. This allows for fast computations, making the method well-suited for scenarios with multiple dynamic obstacles. The framework also extends traditional path planning and trajectory optimization techniques to handle dynamic collisions effectively. A key feature is its ability to detect and respond to hazardous areas created by blind spots, enabling proactive avoidance of potential collisions with unseen dynamic obstacles. Both simulations and real-world experiments validate the effectiveness of our approach, demonstrating its ability to navigate complex and dynamic environments efficiently and safely.