Sadafumi Saito

Measurement of Growth Information of a Strawberry Plant Using a Natural Interaction Device

This paper details a measurement procedure for acquiring growth information on strawberry plants, such as height, width, volume, leaf area, leaf inclination and leaf color using a relatively inexpensive natural interaction device (NID). An NID is a motion-sensing device that is mainly used as a game controller, for example, Microsoft’s Kinect sensor, that can detect the movements of players by means of a color camera and depth sensor. It captures RGB images of 640 x 480 resolution along with per-pixel depth information. We installed an NID on a circulating-type movable bench system for strawberry cultivation. Since all the plants pass under the NID during the cultivation system, color and depth images of all the plants can be obtained. Using an image-processing method based on grayscale matching, color and depth images of every bench were smoothly tiled. Whole-leaf regions for every bench were then derived. To analyze the color and depth images in the whole-leaf regions efficiently, image-processing software was developed to extract and map the growth information on strawberry plants on the movable bench system. To evaluate the accuracy of our measurement procedure, we compared the actual data on height, width, and leaf area with their estimated values. The results showed each coefficient of determination to exceed 0.8. The maps created revealed significantly local biases of plant growth that varied according to strawberry cultivar.

Development of Robotic Strawberry Harvester to Approach Target Fruit from Hanging Bench Side

Abstract To reduce the labor involved in strawberry production, we have developed a strawberry harvesting robot that can operate on a hanging bench culture. In the hanging bench culture, a plant is grown on a bench that has been hung from a beam of the greenhouse. The bench is set at a height of approximately 1.0 m from the floor. In general, mature fruits are located behind immature fruits as viewed from the aisle side. In such a case, a conventional robotic harvester tends to fail in approaching the target fruit because of obstacles such as neighboring immature fruits. As viewed from the hanging bench side, mature fruits are located in front of the immature fruits. Considering this case, our robot has a unique feature that it approaches the target fruit from the hanging bench side and positions itself right below the hanging bench. We have developed a traveling platform for the robot to traverse on and accomplish complete harvesting tasks in a 2D area in a greenhouse. The traveling platform has a gantry structure that moves on a set of three rails. The performance of the traveling platform has been tested, and basic stable movements have been observed in the transverse and the path direction. A negative longitudinal slippage has been observed when a weight of 200 kg was loaded. Basic performance tests of the robotic harvester have confirmed that the robot can successfully harvest mature fruits.

Structural Environment Suited to the Operation of a Strawberry-harvesting Robot Mounted on a Travelling Platform

To establish a structural environment suited to the operation of a strawberry-harvesting robot, a rolling-type hanging bench and travelling platform were developed. The rolling-type hanging bench, consisting a bench frame, a driving shaft, a DC motor, a pinion-rack and a rolling plate, moved left and right along the beam of the greenhouse and could change the width of the path. The travelling platform, consisting of a main frame and a table, was a gantry structure that enabled a robot to move both in the path direction and sideways. No severe mechanical vibration during stepwise path travel at speed of 182 mm s−1 was observed, although negative slippage was occurred. It was verified that the travelling platform performs stable travelling.

Development of a Movable Bench System for Strawberries and the Potential for Plant Growth Measurement

Abstract A circulating-type movable bench system was installed in a research greenhouse in Miyagi Prefecture. The system is 13.5 m long and 7.7 m wide, and chiefly comprises two longitudinal conveying units, two lateral conveying units, two nutrient supply units, a chemical sprayer, 52 planting benches, and a control unit. The cycle times for each speed mode of lateral convey mode were 44.5 s at low speed, 28.8 s at medium speed, and 23.8 s at high speed. The waiting time can be shortened in high speed mode. A prototype of a plant growth measurement system was then developed in a laboratory to estimate plant height and width using a motion control sensor to capture an RGB image and depth image when the bench passed it. Our results suggest fixed-position observation using the movable bench system to be feasible, although further work will be needed to clarify the accuracy of measurement and the relationship between analysed image data and plant condition.