Green shoot thinning (removing unnecessary shoots to redirect vines energy to most significant shoots) in vineyards is used to reduce the crop load to achieve higher flavor concentration in wine grapes. Mechanical green shoot thinning reduces the labor usage by ~25 times compared to manual green shoot thinning. However, due to difficulty in precise placement of the thinning end-effector to the trajectories of cordons, cluster removal efficiencies of green shoots vary between 10-85%. Automating the mechanical thinning operation could help to substantially increase its efficiency and performance. For the automated thinning operation, first step is to develop a machine vision system that can determine cordon trajectories during thinning season in real-field conditions. However, during thinning season growth of shoots/leaves occlude significant portion of cordons making it challenging to accurately determine their trajectories.
The focus of this research is on the study and evaluation of a machine vision system and integrated prototype for automated green shoot thinning in vineyards. A deep learning-based machine vision system was designed to estimate the cordon trajectories from different growth stages of green shoots under real vineyard environment. The proposed approach considers the location information of visible segments of trunk, cordon and density of shoots/leaves to accurately estimate the cordon trajectories in full foliage canopies during varying growth stages of green shoots. This deep learning-based approach helped to estimate cordon trajectories with high correlation coefficient of 0.997, 0.996, and 0.991 from different growth stages of green shoots (week 2 through week 4). The integrated automated green shoot thinning prototype with a low cost RGB-D (red, green, blue, and depth) camera can precisely position the thinning end-effector to the estimated cordon trajectories with an Root Mean Squared Error (RMSE) of 1.47 cm at forward speed of 6.6 cm.s-1(0.24 km.h-1) with averaged initial processing time of 8 s for a single cordon. The results from this study showed the potential of machine vision-based automated green shoot thinning operation in vineyards.
|Advisor:||Zhang, Qin, Karkee, Manoj|
|Commitee:||Whiting, Matthew D.|
|School:||Washington State University|
|Department:||Biological and Agricultural Engineering|
|School Location:||United States -- Washington|
|Source:||DAI-B 82/3(E), Dissertation Abstracts International|
|Subjects:||Agricultural engineering, Robotics, Artificial intelligence|
|Keywords:||Artificial intelligence in agriculture, Automated green shoot thinning, Automation in vineyards, Deep learning, Grapevine canopy detection, Machine vision in agriculture|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be