To speed up soybean seeding, we have been developing a ridge-forming seeder with disc cultivation and a double plate-type seeding mechanism. With an experimental machine, a maximum seeding speed of up to 1.5 m/s was achieved. However, when chemical-coated seeds were used, 20 % were damaged and 30 % were missed. Those chemicals were insecticide and fungicide, and the utilization of those seeds has increased in recent years. The dynamic friction resistance of the seeds was doubled by the coating; moreover, it reached was tripled when chemicals were adhered to the inside of the seeder. To solve these problems, a U-shaped pocket seeding plate was proposed. After the tests of four brands and three sizes, damaged seeds were almost eliminated. In addition, stable operation was achieved to within the 2 % of missing seeds, and the release speed was 20 /s (equivalent to a working speed of 2.5 m/s).

The precise estimation of fish volume is important for monitoring fish growth. In this study, we fabricated and utilised an underwater double cavity Helmholtz resonator consisting of a generating cavity and a portable light-weight acrylic sample cavity for fish volume estimation. A sinusoidal chirp wave with a frequency range of 100 to 500 Hz was generated and captured using an underwater speaker attached to the generating cavity. As the cumulative volume of both the Japanese rice fish (Oryzias latipes) and the Japanese dace fish (Tribolodon hakonensis) increased in the sample cavity, resonance frequency decreased. For both fish species, a high linear correlation with an R² of 0.99 between the actual and predicted volume was obtained.

This study proposes a method of applying image processing on rice plants to determine their status with regard to lodging, which is a contributing factor to declining harvest efficiency. Test images were obtained from photographs of a rice field captured by CCD cameras placed inside cabins, immediately prior to harvesting. A total of 29,870 images were collected for training and evaluation. The image processing system used a histogram of oriented gradients based on each pixel edge to generate a Bag of Visual Words (100-dimensional) by K-means clustering before performing principal component analysis to obtain a 90-dimensional set. A support vector machine was applied to the lodging diagnostic results, and the diagnostic accuracy rate was found to be 74.0 %.

In this study, an image processing system is developed for application on rice plants to determine their status with regard to lodging, which is a contributing factor to declining harvest efficiency and rice grain quality. Test images were obtained from photographs of a rice field captured by CCD cameras placed inside cabins, immediately prior to harvesting. A total of 2,601 images were collected for training and evaluation. A convolutional neural network and an order prediction model using the learning to rank algorithm, RankNet, were used. The normalized discounted cumulative gain (nDCG) was used to assess the quality of the model. The image processing results yielded nDCG@101 was 1.0 and showed that the system was usable as a lodging diagnostic tool.

This study proposes a distance sensor system for a robotic mower by combining object detection with deep learning and distance detection with stereo camera. Herein, a human subject in grass was used as a detection object and the sensor system was able to acquire the distance and direction of the target, even when it was partially covered. The grass had a height of 1 m, detection was stable up to distance of 7 m, and error was 0.2 m at distance of 4 m and 0.1 m at that of 7 m. The error of the direction measurement increased with the increase in the angle from the camera. The maximum error was 4.3° at 46°.

To reduce spray drift for trellis-trained orchard crops, we manufactured an experimental orchard sprayer to use spray arms that could spray closer to trellis, with flat fan nozzles. In a vineyard, a drift distance was reduced from 30 m at spraying by the conventional orchard sprayer with a fan’s air volume of 4.8 m³/s to 10 m with either no air blast or an air volume of 3.2 m³/s at spraying by the orchard experimental sprayer. In a Japanese pear orchard, it is not clear that drift was reduced when the air volume was reduced from 4.8 m³/s with the conventional orchard sprayer to 3.2 m³/s with the experimental orchard sprayer.

At present, the expansion of green soybean cultivation is advancing in Akita prefecture, but mulch-covering is necessary for early cultivation and mechanization of seeding work is delayed. Therefore, in this study, we examined the simultaneous work with a surface fine earth rotary and mulch seeding machine. Each simultaneous operation was satisfactory. The pulverizing rate was 90 % or more and a high emergence rate was obtained by mulch. The emergence position was within 1.7 cm on average from the center. Finally, it was estimated that a good product yield equivalent to that of conventional cultivation could be obtained and the maximum burden area could be expanded by 6.3 times with high efficiency and labor saving.

A rice transplanter designed for high-density seedlings can pick up small seedling blocks from seedling mats. We compared the planting accuracy of high-density, nursling, and young seedlings using a rice transplanter designed specifically for high-density seedlings. The rice transplanter planted approximately four plants per hill for all three types of seedlings by adjusting the slide times of the seedling mats and the lengths of the seedling blocks. High-density and young seedlings were transplanted from soil in a tray, 74 %-87 % of the picked seedlings were planted at a rate of 2 to 6 plants per hill. Because of the light weight of the seedling blocks with higher density or nursling seedlings, the missing hill rate exceeded 5 % in some cases.