Agile Obstacle Avoidance Control Based on Dynamic Velocity Obstacle Perception

To enhance the obstacle avoidance capability of quadrotor UAVs for high-speed dynamic obstacles, this paper proposes a Dynamic Velocity Obstacle-aware Model Predictive Control (DVO-MPC) method. The proposed method integrates trajectory prediction of dynamic obstacles with a velocity obstacle discrimination mechanism to identify potential collision risks in a multi-step prediction horizon in real time, generating the shortest-time avoidance trajectories oriented toward the target velocity direction. A model predictive control strategy is employed to locally optimize both the avoidance and original mission trajectories, while a differential flatness-based controller ensures rapid attitude tracking. This design enhances the overall responsiveness and robustness of the system, enabling agile and reliable obstacle avoidance in complex, dynamic environments. The effectiveness of the proposed DVO-MPC is validated through high-fidelity software-in-the-loop simulations, which demonstrate superior performance in both hovering and flighting scenarios, as well as agile control under high-speed flight conditions. Comparative results show that the proposed method significantly outperforms conventional algorithms.Potential application scenarios include urban search and rescue, low-altitude dynamic traffic, and evasion in special mission environments.