To realize a mowing robot for levees of agricultural fields, an autonomous driving control system was developed using a low-cost single-frequency real-time kinematic Global Navigation Satellite System (RTK-GNSS) receiver and a low-cost Inertia Measurement Unit (IMU). The difference of observation values between a single-frequency RTK-GNSS receiver and a dual-frequency one was 0.8 cm RMS even while the mower was running. As a result of autonomous running test, the lateral and heading errors were 2.5 cm RMS and 2.2° RMS respectively. The robot mower was capable of running within the target lateral error of 10 cm and the effectiveness of the autonomous controller developed in this research was confirmed. The speed of the mower was 0.54 m/s on average and the workable area per one hour was 770 m².

We conducted basic experiments using a vertical cylinder to assess the possibility of estimating the percentage of ripened grains (PRG) based on the bulk density of rough rice piled in the combine grain tank. Regression lines between the bulk density and particle density (an indicator of PRG) of rough rice piled in a vertical cylinder showed high coefficients of determination (0.71-0.96). Meanwhile, the regression lines shifted toward larger bulk density as the filling condition became denser. The slope of the regression lines changed with the moisture content. The results show that the filling conditions in a grain tank and the effects of moisture content on the regression lines need to be further clarified to estimate PRG using bulk density.

Deteriorated sugarcane leads to decreased quality and lower efficiency in the sugar crystallization process. In this study, we investigated a quality evaluation method for shredded samples using near infrared spectroscopy (NIRS). Most of the deteriorated samples could not be detected with polarimetric method. Therefore, the Brix value was used for a quality index of fresh and deteriorated cane in this test. A calibration model developed with fresh samples gave lower accuracy for deteriorated samples (RMSEP＝0.82 °Brix). However, a calibration model developed with fresh and deteriorated samples was more accurate than the previous test (RMSEP＝0.49 °Brix). We concluded that NIRS is a useful method to evaluate quality of deteriorated cane samples.

We developed an adjoint working method of harrowing and seeding using GPS automated steering system, and evaluated it. Using this method, we could harrow soil accurately based on GPS location information, and seed by manned tractor simultaneously. In a demonstration of this method, the working time in a field of 5 ha was reduced by 34.9 % compared with the conventional method. With this method, seeding, which required a high driving technology in the past, becomes easier. A person with intermediate skills can work with a precision that is equal to that of an expert. This method is also adopted for the robot tractor, which is being developed by manufacturers in Japan.

With regard to a management boom sprayer, three types of high-rigidity booms that reduce the displacements in both the vertical and the horizontal direction were developed, in order to verify their performances in the obstacle driving test (a both-side-wheels riding test and one-side-wheel riding test) and the heavy braking test. In the both-side-wheels riding test, the truss boom (tapered) reduced the vertical displacement (53 %), and all high rigidity booms did not reduce the vertical displacement in a right-side-wheel riding test. Moreover, in the heavy braking test, the high-rigidity slide boom reduced the horizontal displacement substantially (16 %), the truss boom (tapered) reduced the convergence time of horizontal displacement most (63 %).