Abstract
According to the crop-row detecting method which had been developed and analyzed in Part 1, a crop-row detector was made on trial basis. The time required to detect a crop-row by this detector was about 41ms. This is short enough for practical use. As a result of the stationary test that was conducted in actual fields, the offset to a row of crop can be known up to ±10cm and heading up to ±6° with practical adduracy, although the detectable limits was restricted in comparison with simulations. Adoption of a dichroic detection for easy extraction of crop-row will improve the practicability. However, averaging might be necessary for the row of grown-up crops, as the standard deviation of detected offsets was more than 2cm.
