ARDEE: A general agricultural robotic development and evaluation environment

Young, Hunter

Permalink

https://hdl.handle.net/2142/104729 Copy

Description

Title

ARDEE: A general agricultural robotic development and evaluation environment

Author(s)

Young, Hunter

Issue Date

2019-01-14

Director of Research (if dissertation) or Advisor (if thesis)

Chowdhary, Girish

Committee Member(s)

• Grift, Tony E.
• Rodriguez, Luis F.

Department of Study

Engineering Administration

Discipline

Agricultural & Biological Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• field robotics
• development environment
• hardware-in-the-loop
• ROS
• Gazebo
• 3D simulation

Abstract

When evaluating any algorithm, it is essential that the data used for evaluation be collected from the target operating environment, as well as conditions, in order to get an accurate representation of the algorithm's performance in that environment. This is especially important when extrinsic sensor measurements are used for developing and evaluating autonomous control and perception algorithms intended for agricultural applications. Unfortunately, there are many obstacles that can considerably hinder the development process, most notably the 7-8 months in which most crops are not in season. The work presented in this thesis allows the year-round development and evaluation for a wide variety of autonomous control and perception algorithms for agricultural field robotic applications, using a set of developed simulation tools in combination with an open-source simulation platform, Gazebo. The custom set of tools was designed such that any number of user-specific agricultural environments can be simulated, and the sensor/robot configuration can be easily customized, being useful for a wide range of agricultural research interests. The fundamental contributions of this work are the following: (1) a collection of sufficiently accurate simulated crop models for three different crop species (corn, sorghum, and tobacco), (2) user-friendly tools for generating a user-customizable agricultural field environment, (3) a collection of simulated, commonly-used, sensors that can be attached to any simulated robot platform, (4) a simulated model of an ultra-compact robot platform, and (5) a set of socket-based, or UDP, tools used for testing algorithm performance on-board target hardware and with the simulated sensors and field. Finally, a few core autonomous control and perception algorithms, which reflect the range of field robot research areas that could be used, are executed on-board an ultra-lightweight ground robot platform, and the performance is evaluated and compared in both a real-world and a simulated, agricultural environment.

Graduation Semester

2019-05

Type of Resource

text

Permalink

http://hdl.handle.net/2142/104729

Copyright and License Information

Copyright 2019 Hunter Young

Owning Collections