Abstract

In recent years, the interests of introducing autonomous robots by growers into agriculture fields are rejuvenated due to the ever-increasing labor cost and the recent declining numbers of seasonal workers. The utilization of customized, autonomous agricultural robots has a profound impact on future orchard operations by providing low cost, meticulous inspection. Different sensors have been proven proficient in agrarian navigation including the likes of GPS, inertial, magnetic, rotary encoding, time of flight as well as vision. To compensate for anticipated disturbances, variances and constraints contingent to the outdoor semi-structured environment, a differential style drive vehicle will be implemented as an easily controllable system to conduct tasks such as imaging and sampling. In order to verify the motion control of a robot, custom-designed for strawberry fields, the task is separated into multiple phases to manage the over-bed and cross-bed operation needs. In particular, during the cross-bed segment an elevated strawberry bed will provide distance references utilized in a logic filter and tuned PID algorithm for safe and efficient travel. Due to the significant sources of uncertainty such as wheel slip and the vehicle model, nonlinear robust controllers are designed for the cross-bed motion, purely relying on vision feedback. A simple image filter algorithm was developed for strawberry row detection, in which pixels corresponding to the bed center will be tracked while the vehicle is in controlled motion. This incorporated derivation and formulation of a bounded uncertainty parameter that will be employed in the nonlinear control. Simulation of the entire system was subsequently completed to ensure the control capability before successful validation in multiple commercial farms. It is anticipated that with the developed algorithms the authentication of fully autonomous robotic systems functioning in agricultural crops will provide heightened efficiency of needed costly services; scouting, disease detection, collection, and distribution.

Graduation Date

2018

Semester

Spring

Advisor

Xu, Yunjun

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Mechanical and Aerospace Engineering

Degree Program

Mechanical Engineering

Format

application/pdf

Identifier

CFE0007401

URL

http://purl.fcla.edu/fcla/etd/CFE0007401

Language

English

Release Date

November 2021

Length of Campus-only Access

3 years

Access Status

Doctoral Dissertation (Campus-only Access)

Share

COinS