Robust Real-Time UAV Target Tracking with Onboard Vision-Based Yaw Control

Authors

• Rylan Malarchick, Embry-Riddle Aeronautical UniversityFollow
• Jose Castelblanco, Embry-Riddle Aeronautical UniversityFollow
• Enrique Amaya, Embry-Riddle Aeronautical UniversityFollow
• Carmen DiMario, Embry-Riddle Aeronautical UniversityFollow
• Graysen Brinkman, Embry-Riddle Aeronautical UniversityFollow
• Chirag Kumar, Embry-Riddle Aeronautical UniversityFollow
• Kiwon Yoon, Embry-Riddle Aeronautical UniversityFollow
• Sajid Berhane, Embry-Riddle Aeronautical UniversityFollow

Faculty Mentor

Dr. Sergey Drakunov

Abstract

Autonomous tracking of agile unmanned aerial vehicles (UAVs) presents significant challenges for real-time perception and control systems. This work presents AIRHOUND (Autonomous Intelligent Rotorcraft for Hostile Object Unified Navigation and Detection), a UAV platform implementing vision-based yaw tracking through a modular ROS2 software architecture. The system employs YOLOv8 object detection optimized with NVIDIA TensorRT for embedded deployment on an NVIDIA Jetson Orin companion computer. Detected targets are processed through a geometric tracking module that converts pixel coordinates to angular yaw errors using pinhole camera intrinsics, with a proportional controller generating rate-limited yaw commands. These commands are streamed to a PX4 flight controller via the Micro XRCE-DDS bridge for offboard control execution. The software architecture was validated through a 600-second end-to-end integration test in the PX4 Software-In-The-Loop (SITL) simulation environment. Results demonstrated sustained 30 Hz message throughput across all pipeline stages with zero dropped messages, exceeding the 15 Hz minimum requirement by a factor of two. The integration process identified and resolved seven software bugs prior to successful validation, demonstrating the value of simulation-based testing before hardware deployment. This work establishes a validated software foundation for autonomous UAV tracking, with the modular three-node architecture enabling independent development and testing of perception, tracking, and control components. Future work includes hardware validation with real YOLOv8 inference on the assembled platform, transition to transformer-based detection models (RF-DETR), and implementation of Kalman filtering for improved target dropout handling.

Recommended Citation

Malarchick, Rylan; Castelblanco, Jose; Amaya, Enrique; DiMario, Carmen; Brinkman, Graysen; Kumar, Chirag; Yoon, Kiwon; and Berhane, Sajid (2025) "Robust Real-Time UAV Target Tracking with Onboard Vision-Based Yaw Control," Beyond: Undergraduate Research Journal: Vol. 9 , Article 6.
Available at: https://commons.erau.edu/beyond/vol9/iss1/6