High-speed telescope autofocus for UAV detection and tracking.

Area of Science:

• Optics and Photonics
• Aerospace Engineering
• Computer Vision

Background:

• Unmanned Aerial Vehicles (UAVs) pose increasing challenges for detection and tracking systems.
• Existing telescope-based systems require enhanced autofocus capabilities for dynamic targets.
• Deep learning object detection provides a foundation but needs robust real-time focus adjustment.

Purpose of the Study:

• To develop and evaluate a high-speed automatic focus module for a telescope-based UAV detection and tracking system.
• To integrate linear stages and passive focus algorithms for rapid autofocus adjustment.
• To assess the performance of different focus algorithms and contrast measures for fast-moving UAVs.

Main Methods:

• Implementation of linear stages and passive focus algorithms integrated with a dual-telescope system.
• Utilizing deep learning for initial UAV detection.
• Experimental evaluation of Tenengrad operator and Hill Climbing search function for autofocus.
• Field testing with UAVs at varying speeds and distances.

Main Results:

• Successful tracking and continuous focus maintenance on UAVs flying at speeds up to 24 m/s.
• Effective focus range demonstrated from over 4500 m down to 150 m.
• The Tenengrad operator combined with the Hill Climbing search function showed optimal performance for focusing on small, fast-moving UAVs.

Conclusions:

• The proposed high-speed automatic focus module significantly improves the capability of telescope-based systems for UAV detection and tracking.
• The system demonstrates robust performance under dynamic conditions, maintaining focus on high-speed targets.
• Specific algorithms (Tenengrad and Hill Climbing) are identified as highly effective for this demanding application.