Main content

Simplicity Driven Solution for Recognition and Localization of Tomatoes and Fruits in a Robotic Harvesting Environment

Show full item record

Title: Simplicity Driven Solution for Recognition and Localization of Tomatoes and Fruits in a Robotic Harvesting Environment
Author: Chen, Xuming
Department: School of Engineering
Program: Engineering
Advisor: Simon, X. Yang
Abstract: Harvesting robots have been increasingly used in greenhouses and orchards in Canada, USA, and other countries. Tomatoes are the most popular greenhouse product, and oranges, apricots and plums are popular fruits in orchards. For better tomato and fruit freshness and automatic harvesting capability, real-time recognition and localization of tomatoes and fruits are becoming critical. The robot vision system is the most important aspect of the process for detecting and recognizing the objects upon which a harvesting robot works. This research aims to develop a more practical robot vision system that can be easily implemented in the real world and is computationally efficient. First, a single color component (V or U in a YUV color space) is used to segment out the region of interest (ROI) of the harvest objects from the harvest background, stems, and foliage. Second, an edge extractor with constrained curvature is developed to extract the arc edges within a certain radius range based on the model (definition) of the harvest objects. Based on the qualified edges, the central points of each qualified harvest object in the ROI are calculated. Finally, according to the circle-to-circle relationship found in the geometric analysis, the overlapping area between two harvest objects is calculated, and the depth ordering of the two overlapping harvest objects is determined. In the further research, a sub-block object searching algorithm is proposed to solve the problem of the missing objects’ boundaries. For advanced harvesting requirements of online sorting, a salience-based algorithm is proposed to discriminate the degrees of ripeness in tomatoes. For computing acceleration, task scheduling and load balancing are conducted for multi-core and cloud-based computing. The key contributions of this thesis are the practical study of multiple aspects of robotic vision problems encountered in the harvesting of tomatoes and fruits.
Date: 2015-11
Rights: Attribution-NonCommercial-NoDerivs 2.5 Canada

Files in this item

Files Size Format View
Chen_Xuming_201601_PhD.pdf 2.702Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record

Attribution-NonCommercial-NoDerivs 2.5 Canada Except where otherwise noted, this item's license is described as Attribution-NonCommercial-NoDerivs 2.5 Canada