College - Author 1

College of Engineering

Department - Author 1

Electrical Engineering Department

Degree Name - Author 1

BS in Electrical Engineering

College - Author 2

College of Engineering

Department - Author 2

Electrical Engineering Department

Degree - Author 2

BS in Electrical Engineering

College - Author 3

College of Engineering

Department - Author 3

Electrical Engineering Department

Degree - Author 3

BS in Electrical Engineering

Date

6-2026

Primary Advisor

Siavash Farzan, College of Engineering, Electrical Engineering Department

Abstract/Summary

An increase in agricultural demands and the need to sustainably protect crops has created a need for autonomous navigation through a vineyard with a verified pest detection and deterrence system. Developing this solution revealed several challenges. First, the robot must reliably navigate through structured, repetitive vineyard rows. Second, pests must be detected accurately under changing environmental conditions. Lastly, the system must respond with a chosen deterrent within an appropriate time window, all while minimizing false triggers. To address these challenges, this project uses a Husky A200 unmanned ground vehicle equipped with a 2D LiDAR sensor, an Intel RealSense RGB camera, and an onboard computer running ROS 2. The navigation pipeline was first developed using SLAM, AMCL localization, and Nav2 waypoint navigation. This approach worked suc- cessfully in structured indoor environments where walls, corners, and obstacles provided distinct LiDAR features. However, vineyard testing revealed that repeated vineyard row geometry caused row-aliasing issues, where the robot’s estimated pose could jump to a neighboring row. This made static-map localization and pure waypoint navigation less re- liable for field operation. The final approach therefore used a hybrid LiDAR row-following strategy, which enabled the robot to traverse four vineyard rows mostly autonomously. Working in parallel with the navigation pipeline, a YOLO-based object detection system identifies vineyard pests, including deer, squirrels, voles, and birds, using the RealSense camera data. Detections are filtered using confidence thresholds and temporal filtering to reduce false positives. Upon a valid detection, the system autonomously triggers a speaker that plays predator sounds, while an ultrasonic speaker with adjustable settings runs continuously. Field detection performance varied by pest class and distance, per- forming best for larger, well-defined targets at close-to-medium range. The end-to-end system demonstrates the integration of autonomous navigation and robotic vision for real-time pest detection and deterrence, and field testing highlighted both the promise of the hybrid navigation approach and the importance of adapting to dynamic outdoor conditions. This project demonstrates the interdisciplinary nature of robotics in enabling an intelligent and adaptable way of managing vineyard environments.

Download

Share

COinS