Thesis Format
Integrated Article
Degree
Doctor of Philosophy
Program
Electrical and Computer Engineering
Supervisor
Mehrdad R. Kermani
Abstract
Precision agriculture involves leveraging technology to precisely manage farming practices, optimizing crop yields while minimizing resource inputs and environmental impact. Robotic systems play a pivotal role in precision agriculture by automating various tasks such as planting, monitoring crops, irrigation, and harvesting. Equipped with vision and artificial intelligence, these robots gather data and make real-time adjustments, enabling more efficient and sustainable farming practices. The integration of robotic systems in agriculture enhances productivity, reduces costs, and facilitates more precise resource management, ultimately contributing to food security and environmental sustainability.
Propagation facilities and greenhouses play integral roles in modern agriculture by providing controlled environments where seedlings can be carefully nurtured and cultivated under optimal conditions. Propagation facilities employ specialized methods to ensure the successful germination of large quantities of seedlings each year, numbering in the hundreds of millions. These seedlings are nurtured to a specific level of maturity before being packaged and transported to greenhouses, vertical farms, and hydroponic farms. This process ensures a steady supply of healthy and robust plants for further cultivation and eventual harvest, supporting agricultural production and food supply chains. Once the seedlings are deemed ready for shipment, the main stem of the plant is supported by a wooden stake using clips to prevent damage to the plant during transportation. These clips are also used to support the plant initially and later to force it to grow in the desired direction. Similar clips are used in greenhouses to expedite growth and streamline harvesting, while also supporting the plant’s heavy weight during growth and fruiting. Traditionally, the execution of this task depends on the expertise of trained horticultural workers. The stem-stake coupling (clipping) is a labor-intensive process that entails significant manual efforts, both cognitive and tactile, on a large volume of seedlings in these large propagation facilities. The considerable number of seedlings processed annually and the significant manual manipulation involved in this process highlight the importance and benefits of developing a robotic solution for automated stem-stake coupling of seedlings.
This study presents the robotic stem-stake coupling system as an innovative technology to automate agricultural processes, particularly in the realm of seedling cultivation in propagation facilities. The robotic system is designed to emulate human perception in vision and cognition for identifying the suitable stem-stake coupling points and to replicate human motor skills in the dexterity of manipulation for attaching the clip to the stem at the identified clipping point and coupling the stem to a wooden stake.
The machine vision system relies on image processing and machine learning algorithms to identify the plant, the wooden stake, and ultimately an optimal stem-stake coupling point from captured images. Additionally, it computes the actual 3D spatial coordinates of the recognized point from stereo images. The vision system uses the adaptive feature-based object recognition method using feature descriptors to identify various parts of seedlings and plants. The accurate results from this identification enable other autonomous tasks required in suitable stem-stake coupling point recognition. Real-time point recognition using kernel density estimators and real-time point localization using feature-based soft margin SVM-PCA methods are used for identifying the suitable stem-stake coupling points. A comparison between these two methods and other methods based on deep learning approaches is discussed. The proposed algorithms are evaluated using real-world image data from propagation facilities and greenhouses, and the results are verified by expert farmers. The precise outcomes obtained through these identification methods facilitate the execution of other autonomous functions essential in precision agriculture and horticulture.
The mechatronics unit consists of a gantry system with one or multiple robotic arms, each carrying the automatic stem-stake coupling device. The design and structure of the robotic system, along with the architecture of the developed software, adhere to object-oriented principles. This design approach facilitates the expansion of the system by incorporating additional gantries and robotic arms without requiring alterations to the software or gantry configuration. These robotic arms feature an automatic stem-stake coupling device, which makes clips as it is being affixed it the stem and stake. The device makes the clip from metallic wire by shaping the wire into a circular clip. To achieve this, an impedance control method regulates the torque and speed of the motor precisely to form the wire into a desired shape. The entire process is governed automatically, yet it also offers the capability for manual control through human input.
Summary for Lay Audience
This study introduces a novel robotic system to automate the agricultural processes in the realm of seedling cultivation in propagation facilities. This innovative robotic system addresses the time-consuming and labor-intensive task of manual stem-stake coupling. With the rising challenges related to the labor shortage and increasing costs, utilizing a robotic solution for automated stem-stake coupling is paramount for greenhouses and propagation facilities to improve efficiency and sustainability in the industry. The system makes and attaches a clip at a specific location along the seedling stem to couple the stem to a wooden stake. The clip adds support to the seedling during growth and avoids damage during transportation. The robotic system utilizes computer vision methods and machine learning algorithms to identify seedlings and recognize the suitable stem-stake coupling point. It involves analyzing images captured by stereo cameras to identify characteristics specific to seedlings of interest. The system utilizes robotic arms equipped with sensors and actuators for precise control and interaction with the environment. These robotic arms feature an automatic coupling device, which makes the clip. The entire process operates automatically but also allows for manual control with human input.
Recommended Citation
Asadi Shirzi, Moteaal, "A Novel Technology for Precision Agriculture: Robotic Stem-Stake Coupling System" (2024). Electronic Thesis and Dissertation Repository. 10308.
https://ir.lib.uwo.ca/etd/10308
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Multimedia Format
windows_media