An efficient approach to near-optimal 3D trajectory design in cluttered environments for multirotor UAVs

This paper presents a framework for 3D trajectory design in cluttered environments for multirotor unmanned aerial vehicles. The proposed method employs C
²
continuous piecewise Bézier curves to generate smooth and obstacle-free trajectories starting from the path waypoints. The construction of the curves requires the definition of two parameters difficult to estimate, in particular when looking for optimal design. Differential evolution optimization is used to estimate their values on a variety of randomly generated paths. Since the optimization process is too demanding for real-time implementation, an approach to find efficiently near-optimal parameters values is proposed and validated on a simulated dataset.