🤖 AI Summary
Multi-rotor UAVs face significant challenges in achieving high-precision visual interception under coupled camera–vehicle motion and active target evasion.
Method: This paper proposes a novel tightly integrated architecture combining image-based visual servoing (IBVS) with proportional navigation guidance (PNG). A delayed Kalman filter (DKF) is introduced for high-frequency, robust 2D target state estimation, and a field-of-view (FOV)-constrained closed-loop controller is designed to ensure IBVS stability.
Contribution/Results: Simulation results demonstrate a circular error probable (CEP) of 0.089 m—72.8% lower than the state-of-the-art IBVS method. Real-world experiments show >80% interception success rate under wind speeds <4 m/s. To the best of our knowledge, this work is the first to achieve synergistic modeling and real-time closed-loop control of IBVS and PNG for dynamic interception tasks, substantially improving the accuracy, robustness, and practicality of visual servoing in complex operational scenarios.
📝 Abstract
Vision-based interception using multicopters equipped strapdown camera is challenging due to camera-motion coupling and evasive targets. This paper proposes a method integrating Image-Based Visual Servoing (IBVS) with proportional navigation guidance (PNG), reducing the multicopter's overload in the final interception phase. It combines smoother trajectories from the IBVS controller with high-frequency target 2D position estimation via a delayed Kalman filter (DKF) to minimize the impact of image processing delays on accuracy. In addition, a field-of-view (FOV) holding controller is designed for stability of the visual servo system. Experimental results show a circular error probability (CEP) of 0.089 m (72.8% lower than the latest relevant IBVS work) in simulations and over 80% interception success under wind conditions below 4 m/s in real world. These results demonstrate the system's potential for precise low-altitude interception of non-cooperative targets.