Graduate Studies, UNL

 

Dissertations and Doctoral Documents from University of Nebraska-Lincoln, 2023–

First Advisor

Joe Luck

Second Advisor

Santosh Pitla

Degree Name

Doctor of Philosophy (Ph.D.)

Committee Members

Guillermo Balboa, Yeyin Shi

Department

Biological Sciences

Date of this Version

2025

Document Type

Dissertation

Citation

A dissertation presented to the faculty of the Graduate College of the University of Nebraska in partial fulfillment of requirements for the degree Doctor of Philosophy (Ph.D.)

Major: Biological Sciences

Under the supervision of Professor

Lincoln, Nebraska, December 2025

Comments

Copyright 2025, the author. Used by permission

Abstract

Nitrogen management in corn agriculture faces economic and environmental challenges, motivating development of precision, site-specific solutions. This dissertation presents AIR-N (AI-Enabled Robotic Nitrogen Management), a novel system that integrates cloud-based artificial intelligence with an autonomous field robot (Flex-Ro) to optimize nitrogen fertilizer application in real time. The research is driven by the need to improve nitrogen use efficiency while reducing leaching and emissions.

Comprehensive architecture (Cloud-E) was designed using Amazon Web Services to ingest multi-modal agronomic data (soil sensors, crop reflectance, weather) and host machine learning models. A Random Forest-based model was trained on historical field data to predict spatially variable nitrogen requirements; it was optimized via AWS SageMaker Neo for deployment on edge hardware. The Flex-Ro robot, equipped with a custom PWM-controlled sprayer system, receives nitrogen prescriptions from Cloud-E and adjusts application rates on-the-go. The sprayer subsystem includes flow sensors and controlled valves enabling feedback control to achieve target rates with high precision.

Field trials at the University of Nebraska’s NTTL Track and ENREEC research farm was conducted to validate system performance. The Cloud-E platform’s recommendations reduced over-application in low-need zones while maintaining yields in high-need areas, demonstrating agronomic efficacy. Real-time control achieved close alignment between prescribed and applied rates, as flowmeter feedback allowed automatic correction of deviations. System latency from cloud decision to actuation was minimal due to edge computing optimizations. Additionally, a generative AI assistant (AgWise LLM) was integrated to automatically summarize field data and provide decision support, illustrating the potential of language models in farm management.

Key contributions of this work include the development of a fully integrated AI robotics platform for nutrient management, demonstration of an edge-deployable machine learning workflow for real-time agricultural decision-making, and successful field validation of autonomous, variable-rate nitrogen application. The results indicate that combining IoT sensing, cloud analytics, and robotics can significantly advance precision agriculture, reducing input waste and environmental impact while sustaining crop productivity.

Advisors: Joe Luck and Santosh Pitla

Download

Share

COinS