It was observed that the calyx tended to stick to the cushioning material of the suction tube and the fruit was shaken by the rising motion of the suction tube after the fruit had been placed on the single-layer tray. We then added a 30-mm pulling motion of the suction device in the longitudinal direction from the hollow of the single-layer tray after the placing motion and confirmed that the fruits were released smoothly.

Collision Discrimination Using Machine Vision

As described by Hayashi et al. (2011), the pickup equipment had been designed for manually supplied fruits.However, in the system described here, the fruits are supplied automatically by

the supply unit. There is,therefore, a risk that some fruits will be too close together to allow one to be picked up without a collision between the suction device and another fruit in spite of the supply unit having been configured to maintain the supply distance described above. To prevent collisions of this nature, the focal length of the camera was changed from 12 to 6 mm to expand the field of vision to include any fruits around the targeted fruit. When the orientations and the positions of all fruits in the acquired image are located, isosceles trapezoids are drawn on the fruit regions based on the geometric information shown in figure 8. If the trapezoid overlaps other fruit regions, the fruit is judged as risking a collision if the unit were to pick it up. The packing unit thus does not approach the fruit until the situation around it has changed.

PERFORMANCE TEST

The automatic packing system was installed in a warehouse near a large-scale strawberry sorting and packing center in Saga Prefecture, one of Japan’s main strawberry-producing areas. The performance tests were conducted from December 2010 to February 2011. Fruits of the cultivar ‘Sagahonoka,’ which had been sorted by a farmer into three grades (small, medium, and large) were tested separately by grade (table 2).

Figure 7. Tilt motion of the packing unit’s suction device.

Table 2. Mean sizes of tested fruits. Performance of the Supply Unit

Fruits were manually distributed in a harvesting container in random orientation without overlap. In each test, three sized groups of fruits were placed in one group per each harvesting container. During the test, the fruit conveyer was kept moving and only stopped when the supply unit put a fruit on it. To evaluate the performance of the supply unit, the process time per fruit was calculated,and the task success rate was derived by counting the number of fruits picked up from the harvesting container and successfully put on the conveyer. The number of attempts made by the unit was also recorded to compute the suction success rate. To assess bruising of the tested fruits,30 small, 24 medium, and 20 large fruits were processed by the unit through the

performance test and stored at ordinary temperatures ranging from 4°C to 21°C at the end of

February 2011. Bruises on the fruits were observed by eye two days after the test.

Performance of the Packing Unit

Fruits were manually supplied to the conveyer at spacing of 0.2 m and in random orientations. The process time per fruit was calculated and the task success rate was derived by counting the number of fruits which were picked up, transported, and successfully placed in the required positions in the shipping trays. The postures of the packed fruits were also evaluated. For the returnable tray, if the apex part of the fruit was not in the hole, the posture of the fruit was categorized as failed. For the single-layer tray, we took a picture of the packed fruits and compared it with the fruits which had been packed by a farmer, specifically for orientation and the distance between the center of the fruit region and the hollow. The fruit region and the calyx region of each packed fruit were extracted manually from the digital photo and these geometric features were calculated. The visibility of the calyx was also evaluated by computing the rate of the calyx region versus the fruit region. During the performance test, since the grading function of the packing unit had been deactivated, the fruits were packed in the preset shipping tray