PhD Thesis Defense: Adam Gronewold

Optional ZOOM LINK

"Advances toward a Robotic Management Vehicle Suited to Nurse Row Crops to More Efficient Outcomes"

Abstract

Agricultural production needs revolutionary approaches to produce more with less. Advancements in computation, imagery, and sensors are rapidly driving agriculture toward more efficient outcomes, and automation technologies are increasing available to manage tasks central to perennial crop development. Automation in row-crop agriculture, by contrast, lags behind. This thesis explores utilizing small, unmanned ground vehicles to revolutionize row cropping through the implementation of unconventional management strategies that are challenging to growers using conventional equipment. 

The first focus of this work considers improvements to nitrogen fertilization using small, autonomous vehicles. An agronomy experiment was performed in corn to study the effects of applying nitrogen fertilizer slowly over time to nurse row-crops toward higher yields while reducing input fertilizer costs. Alongside parametric analysis, the feasibility of using micro-UGVs to effectively manage fertilization, despite their payload limitations, is demonstrated. To unlock these precision practices, however, autonomous operations under-the-canopy must be improved substantially. Navigating densely planted row-cropping environments with cameras or ranging sensors is challenging due to sensor occlusion; lighting variability; and the inability to distinguish flexible, traversable surfaces like leaves from rigid obstacles like stalks. A novel tactile-based perception system comprising a mechanical feeler sensor and supporting algorithms was engineered to detect nearby obstacles like rigid cornstalks while filtering out flexible features like weeds and leaves. 

Then, through simple kinematic relationships, the system was used to accurately determine the position of these obstacles such that a blind robot can traverse the messy environment. Through simulation and real-world testing, robust navigation in complex agricultural conditions is shown, overcoming row curvature, planting gaps, dense weeds, and canopy variability—without relying on vision or ranging sensors. Additionally, mobility is crucial to navigating row crops; tight row spacing constrains vehicles and limits their ability to recover from immobilizing conditions. A second prototype vehicle is presented with innovative mechanical systems that enable mobility restoring control in response to incipient immobilization detection and extradition from immobilizing terrain through nontraditional forms of locomotion. 

The results of this thesis establish novel models and methods to overcome visual-impairment and immobilization for agricultural robots, enabling new, unconventional forms of precision agriculture for row crops.

Thesis Committee

• Laura E. Ray
• Minh Q Phan
• Klaus Keller
• Sarah Masud Preum
• Lance Gibson

Contact

For more information, contact Thayer Registrar at thayer.registrar@dartmouth.edu.