The mechanical and algorithmic design of in-field robotic leaf sampling device

Leaf samples analysis is a significant tool to acquire the actual nutrition information of crops. After that, farmers can adjust fertilization programs to prevent nutritional problems and improve the yield of crops. The traditional way for leaf sampling is manual, and researchers need to go to the field and use paper hole punchers with a catch-tube to collect leaf samples. The temperature in summer is hot, and some crop like corn is difficult for researchers to walk through, therefore the manual way of leaf sampling is not a good option. In this thesis, an automatic method of leaf sampling is presented to solve the difficulty of leaf sampling. The contributions of this thesis are the following: (1) Build the end effector of leaf sampling device to punch and store leaf samples separately, (2) Train a neural network to detect the leaves with high horizontal level, (3) Combine point cloud data from the depth camera and vison data from the camera via the sensor fusion to get the leaf rolling angle and grasp point. The method in this thesis can produce a consistent leaf rolling angle estimate quantitatively and qualitatively on multiple corn leaves, especially on leaves with multiple different angles.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Agricultural & Biological Engr

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2021

Author dc:creator

• Wu, Junzhe

Contributors dc:contributor

• Chowdhary, Girish
• Allen, Cody Michael
• Stasiewicz, Matthew Jon

• Leaf sampling
• End effector
• Neural network
• Sensor fusion.

Statement dc:rights

• Copyright 2021 Junzhe Wu

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/110652

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/110652