State-aware reactive curvature-constrained guidance method for agile UAV navigation in 3D environments

This work proposes a state-aware reactive trajectory generation framework for agile UAVs operating in complex three-dimensional environments. The main contributions of this research are:
(1) a momentum-preserving Bézier trajectory generation approach that utilizes the UAV’s instantaneous flight state, including velocity magnitude and heading, to initialize feasible curved trajectories;
(2) an adaptive hierarchical control-point search strategy that efficiently explores promising trajectory regions while reducing unnecessary candidate evaluations through collision-based elimination;
(3) a curvature-constrained multi-objective evaluation cost function based framework that balances directional consistency, trajectory smoothness, path efficiency, and obstacle clearance to generate safe and efficient flight paths.
The proposed framework aims to provide a computationally efficient alternative for real-time reactive guidance of highly maneuverable autonomous UAVs.