PILOT SCOPE - Southeast Center for Agricultural Health and Injury Prevention - PILOT PROJECTS Non-Invasive Monitoring of Cattle Kinematics and Behavior Using Unmanned Aircraft Systems

SPECIFIC AIMS Effective livestock monitoring is essential for promoting animal welfare and improving herd management. However, most existing approaches rely on direct human–animal contact and animal-borne devices, which can be costly, labor-intensive, and challenging to deploy at scale. Unmanned aircraft systems (UAS) offer a scalable, non-invasive option for capturing high-resolution behavioral and kinematic data without direct animal contact. The long-term goal of this project is to reduce or eliminate the need for animal handling and physical instrumentation by developing the basis for a UAS-based, non-invasive cattle monitoring system. Our central hypothesis is that UAS video, combined with image processing and machine learning methods, can accurately estimate key behavioral and kinematic parameters (e.g., velocity, acceleration, heading) currently measured using GPS collars and other animal-borne sensors. This pilot project will accomplish the following specific aims: Aim 1: Collect UAS Video and Telemetry Dataset of Individual Cows. UAS video footage of individual cows during normal grazing conditions will be collected alongside GPS location and heart rate telemetry. This dataset will provide timesynchronized visual, spatial, and physiological information for developing and validating video-based monitoring methods. Aim 2: Develop Image-Processing and Machine Learning Methods for Behavioral Classification and Kinematic Estimation from UAS Video. Post-flight video data will be processed using image-processing and machine learning techniques to detect, track, and estimate motion parameters such as position, velocity, acceleration, and heading. Temporal synchronization and spatial calibration will be implemented to align video-derived estimates with GPS location and physiological telemetry. Aim 3: Validate Video-Derived Kinematic Estimates Using GPS and Physiological Telemetry. Video-derived kinematic estimates will be compared with GPS collar data to quantify model accuracy and assess the feasibility of using UAS video as a reliable method for livestock monitoring.