Date of Award

Spring 2023

Document Type

Open Access Thesis

Department

Mechanical Engineering

First Advisor

Sourav Banerjee

Abstract

In this thesis a novel vision-based AI driven autonomous ‘StakeBot’ is proposed to serve the agricultural 4.0 industries of the future. In recent years supply, demand and production of vegetables that are harvested from plants with weak stems like bell pepper, tomato, eggplant has been significantly up by several fold. With growing demand and new green-house establishments across the country production of vegetables will require a tremendous number of labors which will be hard to supply soon. To overcome the issues in addition to the labor shortages, automation through robotics will be the only viable solution. Plants like bell peppers require support to avoid contamination of the pepper from touching the ground. Thus, a workforce is needed to place these stakes on several acres of land. Alternatively, as adopted in greenhouses across the world, the stakes are placed at a longer distance and rope or wires are tied between them to create continuous support, which makes the plants grow even taller. This is indeed labor-intensive work. To overcome the labor challenges in this thesis, an AI driven robotic solution is proposed that is capable of placing and removing stakes in the ground. With over 210,000 jobs, agribusiness is South Carolina’s No. 1 industry and will be immediately benefited by the proposed solution. The proposed robot is called the StakeBot. The StakeBot is an autonomous self-driving robot which meets the requirements provided by the local farmers from South Carolina. The uniqueness of the StakeBot is its self-driving capabilities which are made possible with a newly developed vision system capable of successfully driving the StakeBot. This vision system utilizes Stereo depth technology paired with the capability of modern Linux based single board computers. This grants the StakeBot the ability to “see” making it capable of avoiding obstacles and making decisions on how to proceed based on the environment. The StakeBot’s vision system is not unique to the StakeBot alone. The developed algorithms can be intergraded for any robotic system that is designed for navigating through its environment. The vision system has shown great success with an average of about 15 FPS, capable of fast response times for vehicles moving at speeds less than 7 MPH.

Share

COinS