Yutaka Kaizu

Development of a Tractor Navigation System Using Augmented Reality

Abstract We developed an intuitive tractor navigation system using augmented reality (AR) by superimposing a computer-generated virtual three-dimensional (3D) image on a camera image. The 3D image was generated using the tractor position and direction determined by two real-time kinematic global positioning systems and an inertial measurement unit. The positioning accuracy of the AR navigation system was examined experimentally by changing the roll, pitch, and yaw angles of a tractor at rest on an actual field. The positioning errors in the world coordinate system were less than 3 cm within 3 m from the front of the tractor, and less than 3 pixels in the image coordinate system. The refresh frequency of the AR image was 30 Hz and the time taken from image capturing to displaying was within 10 ms.

System for Automatic Separation of Sugar Cane Top

In Japan, mechanical sugarcane harvesting has been introduced in many regions. Although it improves the efficiency of sugarcane harvesting, it also brings the cane tops into sugar factories, which decreases yield and causes economic losses. In a former study, a laser scanning technique to distinguish cane tops from the mechanically harvested raw sugarcane materials was investigated. This study improved the techniques and developed a new pattern analyzing method. A green He-Ne laser (wavelength 543.5 nm, output 4 mW) was used to scan the raw sugarcane materials. The back-reflected light intensity was measured by a light sensor (avalanche photodiode module). The surface conditions of each cane sample can vary in terms of the surface roughness and the substances that are adhered to it. Due to the non-uniform surface conditions of cane tops and cane stalks, the cane top and stalk can be distinguished by analyzing the different patterns of the distribution of the back-reflected light intensity. During the experiment, 40 cane top and 40 cane stalk samples were used. The patterns were first analyzed by kurtosis and Fourier transformation, followed by linear discriminant analysis and setting a threshold number to classify these samples. The results showed that the accuracy for overall samples reached 94.90%, from which the accuracies for cane stalks and cane tops were 91.26% and 98.54%, respectively.

Measurement of Sugarcane Surface Roughness Using Laser

In Japan, mechanical sugarcane harvesting has been introduced in many regions. It improves the efficiency of sugarcane harvesting, but also brings the cane tops into sugar factories, which decreases the yield and causes economic losses. This study investigated a system that can distinguish cane tops from the mechanically harvested raw sugarcane materials. We used a green He-Ne laser (wavelength 543.5 nm, output 4 mW) to scan sugarcane and measured the back-reflected light intensity by a light sensor (avalanche photodiode module). Since the surface roughness is different between cane top and cane stalk, analyzing the different patterns of the distribution of the back-reflected light intensity enables the cane top and stalk to be distinguished. In the experiment, 22 cane tops and 32 cane stalks were used as samples. Using kurtosis as a parameter to analyze the patterns, the percentage of correctly identifying cane tops was 90.9, and that of cane stalks was 71.9, and the percentage of correct answers of all samples was 79.6.

Robotic Separation of Stage IV Micropropagated Sugarcane Shoots

A robot system that continuously divides a clump of micropropagated sugarcane seedlings into individual shoots was developed and tested. Micropropagated sugarcane seedlings are generally called sugarcane mericlone, and a technique to produce these seedlings in large quantities by tissue culture has been developed recently. The mericlone sugarcanes grow thickly and their roots become entangled with each other, thus they form a root ball and it is difficult to separate these seedlings. To solve these problems, two new types of end-effectors were developed. The first one was a continuous shoots picking mechanism (CSPM), which was designed to bring the thickly grown shoots into a line and handle them without damage. The second one was a single shoot separator (SSS), which was designed to pull off individual sugarcane shoots from the clump one by one.