This paper presents an approach for real-time autonomous obstacle avoidance for fixed-wing unmanned aerial vehicles (UAVs) for scenarios in which a UAV is required to stay close to a reference path. A key challenge is rapid trajectory generation around obstacles while accommodating vehicle constraints. A UAV model with nonlinear dynamic constraints provides more natural accommodation of the vehicle’s constraints than a kinematic model with linear constraints. This paper presents a method for using finite horizon model predictive control with a custom solver that offers low solution time. A comparative study of a high-fidelity model and a lower-fidelity counterpart is presented. Using the proposed method, the high-fidelity model provides better trajectories than the lower-fidelity counterpart, despite both having low computational requirement for onboard trajectory generation in an embedded platform.
Real-Time Autonomous Obstacle Avoidance for Fixed-Wing UAVs Using a Dynamic Model
Journal of Aerospace Engineering ; 33 , 4
2020-04-02
Article (Journal)
Electronic Resource
Unknown
End-to-End Learning-Based Obstacle Avoidance for Fixed-Wing UAVs
Springer Verlag | 2023
|