This paper describes a strawberry-harvesting robot, a packing robot, and a movable bench system. The harvesting and packing operations in strawberry production require harder, more time-consuming work compared to other operations such as transplanting and chemical spraying, making automation of these tasks desirable. Since harvesting and packing operation account for half of total working hours, automation of these tasks are strongly desired. First of all, based on the findings of many studies on strawberry-harvesting robots for soil culture and elevated substrate culture, our institute of the Bio-oriented Technology Research Advancement Institution and Shibuya Seiki developed a commercial model of a strawberry-harvesting robot, which is chiefly composed a cylindrical manipulator, machine vision, an end-effector, and traveling platform. The results showed an average 54.9% harvesting success rate, 8.6 s cycle time of picking operation, and 102.5 m/h work efficiency in hanging-type growing beds in an experimental greenhouse. Secondly, a prototype automatic packing robot consisting of a supply unit and a packing unit was developed. The supply unit picks up strawberries from a harvesting container, and the packing unit sucks each fruit from calyx side and locates its orientation into a tray. Performance testing showed that automatic packing had a task success rate of 97.3%, with a process time per fruit of 7.3 s. Thirdly, a movable bench system was developed, which makes planting beds rotate in longitudinal and lateral ways. This system brought high density production and labour saving operation at a fixed position, such as crop maintenance and harvesting. By setting up the main body of a strawberry-harvesting robot on working space, unmanned operation technique was developed and tested in an experimental greenhouse. Field experiments of these new automation technologies were conducted and gave a potential of practical use.
To monitor the growth of strawberry fruits in a movable bench system, we developed an algorithm for estimating the fruit diameter using images from a digital color camera and a time-of-flight sensor, which were equipped under the movable bench system. The shapes of the detected fruit were modified using an aspect ratio calculated based on the distance information. A total of 180 fruits within a wide range of maturity were tested using a measurement system. The RMS error for the fruit diameter was 3.5 mm (11.7 %). The accuracy depends on the level of overlap, which is classified into four patterns: (A) there are no overlapping fruits, (B) a fruit is exposed but overlaps to ripened fruits, (C) a fruit is hidden by immature fruits, and (D) a fruit both overlaps to ripened fruits and is hidden by immature fruits. The RMS errors for these patterns were 2.2 mm (8.3 %), 3.7 mm (10.3 %), 3.5 mm (11.5 %), and 4.9 mm (14.7 %), respectively. This indicates that the measurement accuracy is relatively high even when the target fruit is in a state such as (B) or (C).
A circulating movable bench system for strawberries, in which plants on the benches are moved in both lateral and longitudinal directions, was developed, aiming to increase area productivity and reduce the heating and lighting cost per plant. In the system, all benches pass through a fixed point at a lateral conveyer unit every day to be watered. Therefore, it is believed that the growth information of all plants can be obtained at a high frequency by installing sensors at this fixed point. In this study, a method for counting the number of red fruits and immature green fruits in the movable bench system was proposed. In the measurement system, color images and depth information are obtained from below the strawberry plants. A time-of-flight (TOF) sensor was applied to the system to segment overlapping fruits using the depth information to manage the occlusion of the fruits. As a result, the counting success rate for the red fruit reached 96.8 % with "Amaotome" and 94.7 % with "Benihoppe". Without the depth information, however, this rate reached 90.3 % and 74.3 %, respectively. The counting success rate for the immature fruit was 69.6 % and 71.2 %, respectively.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.