You are on page 1of 9

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/issn/15375110

Research Paper

Gentle handling of strawberries using a suction device


Shigehiko Hayashi a,*, Kohei Takahashi b, Satoshi Yamamoto a, Sadafumi Saito a, Takashi Komeda b
a

Institute of Agricultural Machinery, Bio-oriented Technology Research Advancement Institution, National Agriculture and Food Research Organization, 1-40-2 Nisshin, Kita, Saitama 331-8537, Japan b Shibaura Institute of Technology, 307 Fukasaku, Minuma, Saitama 337-8570, Japan

article info
Article history: Received 16 January 2011 Received in revised form 22 April 2011 Accepted 26 April 2011 Published online 15 June 2011

This study investigated the feasibility of gentle handling strawberries using a suction device. This picking-up method, in which the fruit itself is moved towards the suction device by a suctioning airow, is proposed to prevent damage to the pericarp. The picking-up equipment comprises a Cartesian coordinate manipulator, suction device, machine vision system, belt conveyor, and control unit. The suction device has a tapered tube with an inner diameter of 25 mm, and generates a suction airow of approximately 45 l min1. The machine vision system assesses the orientation of the fruit, and the suction device approaches the fruit along the line of fruit orientation. An investigation of the effective space for suctioning revealed that the smaller the fruit, the larger the effective space. Its height was about equal to or slightly greater than half the fruit diameter; however, the permissible distance in the transverse direction was small. Because the inclination of the suctioned fruit varied considerably, our proposed picking-up method was not always able to hold the fruit in a constant posture. In the approach position (80 from the vertical), the suction device required a suction force more than double that required in the vertical position. In picking-up performance tests, success rates for four cultivars were more than 95% without dropping the fruit at an approach height of 16 mm; however, the rate decreased to 71.9% for the long-tapered Deco rouge at a height of 19 mm. The time required to pick and transfer a fruit was 8.9 s. 2011 IAgrE. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

The strawberry (Fragaria ananassa Duch.), which has gained worldwide popularity, is one of the most difcult crops to handle because of its soft pericarp. In Japan in 2008, 190,700 t were produced under a cultivated area of 6471 ha (MAFF, 2007, 178), mostly for fresh consumption. In forcing culture, strawberry plants are transplanted into greenhouses in September and harvested from November to April the following year. Nearly all operations, especially harvesting, sorting, and packing, are performed manually. To harden

the pericarp, the harvested fruits are precooled in storage before packing. They are sorted using a grading standard (size and shape) that differs according to producing area. The strawberries are methodically laid in plastic boxes in two layers to protect them from vibration during transport to the market and to provide visual appeal. This two-layer packing style is the commonest method used in Japan. The use of single-layer packing methods that employ soft sheets with hollows is, however, becoming common. A number of studies on the mechanisation of strawberry grading have been pursued over the years. A grade

* Corresponding author. Tel./fax: 81 48 654 7137. E-mail address: shigey@affrc.go.jp (S. Hayashi). 1537-5110/$ e see front matter 2011 IAgrE. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.biosystemseng.2011.04.014

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

349

classication technique based on the geometric features of a projected image of the fruit was proposed (Nagata, Bato, Mitarai, Cao, & Kitahara, 2000; Nagata, Kinoshita, Asano, Cao, & Hiyoshi, 1996). Imou, Kaizu, Morita, and Yokoyama (2006) developed a 3D shape measurement system using a volume intersection method. Grading systems for strawberries based on two or three characteristics have also been designed (Bato, Nagata, Cao, Hiyoshi, & Kitahara, 2000; Liming & Yanchao, 2010). Other non-destructive methods based on rmness have been reported. An estimation technique for fruit rmness using hyperspectral imaging has been proposed (Tallada, Nagata, & Kobayashi, 2006), and relationships between strawberry fruit rmness and the content of alcohol-insoluble solids and hydrochloric acidsoluble pectin have been elucidated (Kashiwazaki, Nagasue, Soutome, Nakajima, & Omori, 2007). These previous studies were fundamental ndings on grading but could not be incorporated into the mechanical handling of packing. Simple holders, for example, have been used to transport fruit (Nagata, Kinoshita, Asano, Cao, & Hiyoshi, 1997). As an attempt to achieve mechanical packing, Konya and Omori (2010) proposed a peduncle-grip-type tray and handled the strawberry fruit by grasping its peduncle. Nevertheless, a peduncle-grip-type tray of this type has not yet been accepted by the distribution business. Kitazawa, Ishikawa, Nakamura, and Shiina (2008) also proposed a new packing style in which the fruit is held in a vertical position, and conrmed less damage during transport than seen with the conventional two-layer packing style. However, this method requires manual labour. In a few producing areas, a weightsorting machine or non-destructive sorter that grades according to size, weight, sugar content or acid content (Yamada, Tanaka, & Takada, 2009) is used for practical reasons. The picked fruits are placed individually on a conveyor or a transport pan by workers and then sorted automatically; however, the packing task is done manually. Gentle handling techniques are clearly crucial to achieve the mechanisation of strawberry fruit packing. Pneumatic or electromechanical grippers (e.g. Tedford, 1990) and suction techniques are widely utilised for handling delicate agricultural material. For strawberries, force feedback control using mechanical grippers seems to be acceptable; however, the control system risks being complicated and expensive. It would also be difcult to narrow the space between fruits in a package due to contact of ngers with neighbouring fruits. On the other hand, the suction technique is also utilised to harvest or transport crops, including cherry tomato (Subrata et al., 1998), tomato (Monta et al., 1998), mushroom (Reed & Tillett, 1994), and deciduous fruits (Ishii, Toita, Kondo, & Tahara, 2003). The suction technique has certain features that allow the measurement of positional error for target objects and deal with different sizes of agricultural products, but there is no literature that mentions the application of suction techniques to the handling of strawberry fruit. This study examines the gentle handling of strawberry fruit using suction airow, with the ultimate aim of mechanisation of packing. The suction device is employed to suction the calyx side to avoid damage and ease the pick-up motion from a conveyor, since the calyx side of the fruit is signicantly rmer than the equatorial portion or apex side (Konya,

Omori, & Hayashi, 2007). A machine vision system is also used to detect the calyx side before picking-up. A Cartesian coordinate manipulator with a rotary actuator is used to manoeuvre the end-effector (the suction device), since this type of manipulator shows high-precision positioning, highspeed motion and easy control, and recently, much easier and cheaper (Windows-based) control software has improved the operational capability of the manipulator. The end-effector can also be positioned at arbitrary angles with a stepper motor, allowing the suction device to approach the fruit from any direction. The objectives of this study were: (1) to develop a suction device for handling strawberry fruit; (2) to devise a gentle handling method using the suction device; and (3) to conduct picking-up tests using several cultivars that differ in shape. The study endeavoured to obtain an effective space to suction the fruit by examining the positional data between the fruit and the suction device, such as depth, height, angle, and transverse distance.

2.
2.1.

Materials and methods


Strawberry picking-up setup

The strawberry picking-up equipment is composed of a manipulator, a suction device, a machine vision system, a belt conveyor for fruits, and a control unit as shown in Fig. 1. A block diagram of the system is illustrated in Fig. 2. The Personal Computer (PC) can control all the components except the Light Emitting Diode (LED) light. Although three belt conveyors for trays were incorporated in advance for a further study of sorting, this study did not use them. Moreover, the aim of this mechanical handling technique is to combine it with a largescale sorting system such as a non-destructive sorter (Yamada et al., 2009), so the picking-up equipment is designed for fruits that are manually supplied on a belt conveyor. The manipulator is a Cartesian coordinate type with a rotary actuator, designed specically for high-speed performance and ease of control. The actuator is mounted on a linear slide along the z-axis and rotates around this axis, giving it four

Fig. 1 e Photo of strawberry picking-up equipment.

350

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

PC
Motion control Board

Manipulator (4 DOF) Cartesian coordinate manipulator (x-, y-, and z-axis) Rotary actuator (roll)

Suction device
RS232C

Motor controller

Stepper motor for tilting suction device Solenoid valves (suction and purge)

Relay Vacuum sensor


DIO board

Conveyor for fruit Belt conveyor

Machine vision LED light


IEEE 1394

CCD colour camera

Fig. 2 e Block diagram of strawberry picking-up equipment.

Fig. 4 e Schematic diagram of internal structure of suction device to pick up strawberry fruit.

degrees of freedom (DOFs). Fig. 3 shows a schematic diagram of the manipulator. The ranges and speeds of the three slides are 1300 mm and 240 mm s1 for the x-axis, 600 mm and 180 mm s1 for the y-axis, and 300 mm and 150 mm s1 for the z-axis. The actuator rotates 360 at up to 150  s1, so the suction device can approach a target fruit on the conveyor belt from all directions. The manipulator picks up the fruit, transports it, and places it in a hole in the tray. An air-suction method to handle strawberry fruits is proposed in this study because it can compensate for errors caused by machine vision. As shown in Fig. 4, the developed

suction device is composed of a tapered tube, an ejector, and a pressure sensor. The ejector installed at the bottom generates a suction airow of approximately 45 l min1 when compressed air is supplied at 0.6 MPa. The tip of the tapered tube has an inner diameter of 25 mm and is covered with cushioning material that contacts the pericarp of the strawberry fruit. It can be tilted from 80 to 80 by a stepper motor, with a maximum speed of 100  s1. The pressure sensor checks whether the device has successfully suctioned a fruit. Purge air can be supplied while releasing the fruit.

Fig. 3 e Schematic diagram of manipulator.

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

351

The machine vision system consists of a Charge Coupled Device (CCD) colour camera (1024 768 pixels) and two LED light sources (3.6 W each). The distance between the camera and the conveyor is approximately 500 mm, and the illuminance around the conveyor was 130 Lx. The machine vision software was developed using the Halcon image-processing tool to assess fruit orientation. The conveyor has a 100 mm-wide belt and can supply fruits at a speed of 100 mm s1. Cushioning material is attached to its surface to prevent bruising. In Fig. 3, the conveyor conveys the fruit from right to left. When the fruit arrives under the machine vision system, the conveyor stops on receiving a digital signal from the computer.

fruit orientation. Next, the camera rotated clockwise in steps of 30 up to 330 and the software measured the fruit orientation at each position. The fruit orientation angle as measured by the software was compared with the actual orientation. Forty-seven samples of the cultivar Beni-hoppe with a maturity level of more than 70% were used in this test. Preliminary tests showed the measurement error for fruit orientation to be 5.3 with a standard deviation of 4.2 . This appeared to be caused chiey by misdetection of the calyx; its measured position deviated from the actual position when relatively large calyx leaves were bent or clustered together on one side, even though the red fruit fraction was almost perfectly detected. In other words, much of the measurement error consists of rotational rather than deviational error.

2.2. 2.2.1.

Machine vision system Machine vision algorithm 2.3. Lying angle of strawberry fruit and approach angle of suction device
The angle of approach of the suction device appears to be an important factor for picking-up the strawberry fruit lying on the conveyor. Thus, the inclination of a line connecting the fruit apex and the centre of calyx to the horizontal (conveyor belt) was dened as the lying angle, as shown in Fig. 6. Measurement of 55 samples of the cultivar Beni-hoppe revealed the lying angle to be 21.8 with a standard deviation of 3.4 . Given that the fruit lies at angle of about 20 , an appropriate approach angle was investigated. The suction device was maintained at the position shown in Fig. 6, and the success rate of suction was calculated for 20 samples (average weight: 10.4 g). The approach angle was set to 10 , 20 and 30 , and then the height was set at 15 mm and 20 mm. The relationship between approach angle and success rate is shown in Fig. 7. The success rate was above 70%, except for an approach angle of 30 and a height of 15 mm. Moreover, visual observation revealed that the upper tip of the suction device sometimes pushed the fruit at an approach angle of 20 or 30 , raising concerns about the risk of bruising. We concluded that an approach angle of 10 was adequate, although the optimal approach angle was not clear. The following experiments therefore used an approach angle of 10 to investigate the performance of the suction device.

Detecting the orientation of the calyx and fruit is an essential technique for suctioning the strawberry fruit from the calyx side when the suction device picks it up from the conveyor. The machine vision extracts three parts of a strawberry fruit: the whole fruit, the red fraction, and the calyx fraction. Fig. 5 shows a sample picture of image processing. First, the machine vision captures the colour image of the fruit, and inter-picture operation is applied to obtain the grey images. Second, the whole fruit, red fraction, and calyx fraction are respectively segmented using the following equations. 48R 45G > Tw 2R G B > Tr 2G R > Tc (1) (2) (3)

where R, G are B image frames of the colour camera, and Tw, Tr and Tc are the threshold values for detecting each fraction. Third, the centroids of the each fraction are calculated, and the line connecting the centroids of the calyx and red fruit fractions is dened as the fruits orientation. The positional data (rows and columns on the image frame) are transformed from camera coordinates to manipulator coordinates using a previously calibrated homogeneous transform matrix.

2.2.2.

Measurement accuracy of fruit orientation

The machine vision algorithm was examined in terms of its assessment of fruit orientation. A strawberry fruit was placed manually under the camera to align the fruit vertically in the camera frame. In this position, the actual fruit orientation was regarded as 0 , and the machine vision software measured the

2.4.

Picking-up operation

Fig. 8 shows a owchart of the picking-up operation. The fruit conveyor is started and the fruits are placed manually upon it,

Fig. 5 e Image-processing process: (a) captured image, (b) enhanced R-G grey image, (c) R-G-B grey image, (d) enhanced G-R grey image, and (e) resulting image; arrow indicates orientation of fruit.

352

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

Fig. 6 e Approaching motion of suction device and denition of approach angle, lying angle, height, depth and transverse distance.

after the trays are set. At the same time, the camera starts capturing images. When a fruit comes into an image frame, the conveyor stops moving and the above-described machine vision algorithm is executed. The suction device is managed and positioned at the calyx side in a tilted posture (approach angle: 10 ) as shown in Fig. 6, and is advanced towards the fruit along the line of the fruit orientation as assessed by the algorithm. Preliminary tests found that bruises or abrasions on the pericarp occurred due to surface friction with the conveyor belt when the tip of the device pushed the fruit. To reduce these types of damage, a picking-up method was adopted in which the fruit itself moves towards the tip of the suction device as a result of suction airow: the suction device approaches the fruit slowly until a negative pressure is generated, with the maximum approach depth set at 20 mm. The suction device lifts the fruit, transports it vertically, and deposits it in the hole of the appropriate tray using purge air.
100 90 80

2.5. 2.5.1.

Functional tests Effect of approach direction of suction device

The direction of approach of the suction device affects the success rate. A suction test was therefore conducted to clarify both the effective space and adequate space for suctioning the

Start

Move fruit conveyor

Stop conveyor when fruit in frame

Image processing

Control suction device

Success rate, %

70

Suction and approach fruit


60 50 40 30 20 10 0 10 20 30

No

Successful suction ?
Yes

Transport into tray


(210mm horizontal and 135mm vertical position) Yes

Fruit supply
No

Approach angle,
Fig. 7 e Effect of approach angle on suction success rate: ,, 15 mm high of suction device; B, 20 mm high of suction device.

End
Fig. 8 e Flowchart of picking-up operation.

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

353

fruit: effective space is dened in space and represents a suction success rate of 80.0% or more, and adequate space is represented as a suction success rate of 66.7e80.0%. A fruit was placed in front of the suction device, as shown in Fig. 6, and the suction air volume was increased gradually with the device in a xed position. Subsequently, the suctioning state, i.e., whether the fruit was successfully suctioned or not, was investigated. The depth d between the fruit and the suction device was set at 0, 5, or 10 mm. The height h of the tip of the tapered tube from the conveyor surface was set at 16, 19, 22, or 25 mm. In addition, the transverse distance t was set at 0, 5, or 10 mm. Sample fruits were divided into ve groups based on the real weight of the fruit w (in g): w < 10; 10 w < 20; 20 w < 30; 30 w < 40; and 40 w. There were 15 samples of the strawberry cultivar Beni-hoppe in each group.

and the execution time from image capture to release of the fruit into the tray was measured. In addition, bruises or abrasions on the pericarp of the tested fruit were visually inspected immediately after the test and one week later, after keeping them in cold storage at about 10  C. In this test, the cultivars Natsuakari, Deco rouge, Tochiotome, and Beni-hoppe were used, and the approach height was set at 16 mm. The cultivars Natsuakari and Deco rouge, which have shapes that are distinctively different from the others, were also tested at the approach height of 19 mm. Table 1 shows the mean sizes of tested fruits for the four cultivars. Among the four types of fruit shapedlong-tapered, square, tapered, and rotund (Liming & Yanchao, 2010)dNatsuakari appeared to be rotund, Deco rouge was long-tapered, and the other two were tapered.

2.5.2.

Fruit posture in suction device

The posture of the fruit suctioned by the suction device may vary in the forward-backward or left-right direction. Therefore, the inclination angle of the suctioned fruit was measured to evaluate fruit posture. Photos of the front and side views of a fruit were taken using a digital camera after the suction device had picked it up automatically and stopped in the vertical orientation during transportation. A human observer measured the inclination angle from the photos. Thirty samples each of the cultivars Beni-hoppe, Deco rouge, and Natsuakari were used in this test.

3.
3.1.

Results and discussion


Effective space for suction picking-up

2.5.3.

Transportation ability of suction device

To evaluate the ability of the suction device to transport the strawberry fruit, the minimum negative pressure for holding a fruit was measured with the suction device in two congurations: the vertical orientation and 80 to the vertical (posture while picking-up). The device suctioned a fruit such that the orientation of the fruit coincided with the direction of the suction device. The suctioning air volume was then gradually reduced, and the negative pressure when the fruit detached from the suction device was recorded. Sixty samples of the cultivar Beni-hoppe were used in this test.

2.6.

Picking-up performance test

The picking-up performance test was conducted to evaluate system performance. Fruits were manually placed on the conveyor at equal intervals of approximately 200 mm in random orientations, and the picking-up operation was initiated following the owchart shown in Fig. 8. The success rates of picking-up at different approach heights were investigated,

The success rates with the suction device at different heights, transverse distances, and depths are shown in Table 2. The dark grey cells indicate the effective space representing a suction success rate of 80.0% or more, whereas the light grey cells indicate an adequate space representing a suction success rate of 66.7e80.0%. The optimal height for suctioning a fruit increased roughly in proportion to the weight of the fruit; the smaller the fruit, the larger was the effective space. The optimal height was about equal to or slightly greater than half the fruit diameter. If the fruit was too small, it could not be suctioned, since the wide gap between the fruit and the tip of the device prevented the negative pressure from building up to a sufciently high level to produce suctioning airow. In contrast, when the fruit was too large, it could not be moved by the airow. The results showed that a suction device with an inner diameter of 25 mm could pick up fruits weighing less than 40 g with a high success rate if they were located in a space that was amenable for suctioning. For bigger fruits, a suction device with a bigger inner diameter would be more suitable. In this study, the picking-up equipment was designed to deal with several sizes of fruit to identify an effective space for suction; however, the results suggested the potential for using the parallel output lines seen in existing specialised suction devices. Moreover, good picking-up performance is most likely at a height of 16e19 mm for Beni-hoppe. In addition,

Table 1 e Mean sizes of tested fruits. Cultivar


Natsuakari Deco rouge Tochiotome Beni-hoppe

Fruit number
47 32 50 52

Weighta
14.1 16.5 13.6 13.0 (5.0) (2.9) (3.8) (3.3)

Lengtha: L (mm)
34.4 (5.0) 41.8 (4.8) 35.2 (4.1) 36.0 (3.8)

Diametera: B (mm)
32.4 (3.7) 31.6 (2.6) 31.1 (3.4) 30.6 (2.7)

Ratio: L/B
1.06 1.32 1.13 1.18

Shape categoryb
rotund long-tapered tapered tapered

a Values in parentheses are standard deviations. b Based on visual observation.

354

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

Table 2 e Effect of height, traverse distance, and depth of suction device on suction success rate: cultivar Beni-hoppe.
Weight range: w (g) Weight: Avg. (SD) (g) Diameter: Avg. (SD) (mm) Length: Avg. (SD) (mm) Suction success rate (%) ha (mm) 16 w < 10 7.2 (1.3) 24.0 (1.2) 28.9 (3.6) da (mm) 0 5 100.0 93.3 100.0 93.3 86.7 60.0 86.7 80.0 93.3 73.3 66.7 20.0 20.0 26.7 26.7 6.7 13.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 10 w < 20 13.7 (2.5) 29.2 (2.3) 38.8 (3.5) da (mm) 0 5 100.0 80.0 100.0 86.7 53.3 33.3 100.0 73.3 100.0 66.7 40.0 0.0 93.3 53.3 80.0 46.7 13.3 6.7 26.7 6.7 13.3 13.3 0.0 0.0 20 w < 30 24.1 (2.2) 33.9 (1.5) 48.3 (3.1) da (mm) 0 5 100.0 66.7 60.0 6.7 0.0 6.7 100.0 66.7 73.3 6.7 6.7 0.0 93.3 13.3 46.7 6.7 6.7 0.0 20.0 6.7 6.7 6.7 0.0 0.0 30 w < 40 35.6 (3.4) 38.8 (2.2) 53.5 (3.9) da (mm) 0 5 26.7 6.7 20.0 0.0 0.0 0.0 73.3 13.3 53.3 0.0 13.3 0.0 93.3 6.7 53.3 6.7 6.7 0.0 73.3 0.0 33.3 0.0 0.0 0.0 40 w 46.6 (3.6) 40.4 (2.2) 61.8 (2.9) da (mm) 0 5 0.0 0.0 0.0 0.0 0.0 0.0 6.7 0.0 0.0 0.0 0.0 0.0 6.7 0.0 6.7 0.0 0.0 0.0 13.3 0.0 13.3 0.0 0.0 0.0

ta (mm) 0 5 10 0 5 10 0 5 10 0 5 10

19

22

25

10 20.0 13.3 0.0 6.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

10 13.3 13.3 0.0 13.3 13.3 0.0 6.7 6.7 0.0 6.7 0.0 0.0

10 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

10 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

10 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

a h is the height of the suction device; t, the transverse distance; and d, the depth distance between fruit and suction device illustrated in Fig. 6.

a transverse distance of 5 mm would be permissible for small fruits weighing less than 20 g.

a picking-up method using suctioning airow is not always able to maintain a stable fruit posture.

3.2.

Fruit posture by suction picking-up

3.3.

Ease of transportation

Fig. 9 shows the inclinations of the suctioned fruit viewed from the front and side. The inclinations for Beni-hoppe and Deco rouge varied widely; the maximum angle of inclination was approximately 33 for both views. In contrast, Natsuakari was suctioned at smaller inclinations. For Beni-hoppe and Deco rouge, errors in approach direction caused larger inclinations than in Natsuakari because of the formers elongated shapes. Moreover, these results show that

40

Inclination from side view,

20

Of 60 samples, 12 could not be suctioned if the device was oriented at 80 from the vertical; the weight range of unsuccessfully suctioned fruits was 36e52 g. These failures were caused by the fruit turning in this conguration or due to roughness on the calyx side, both of which led to insufcient air pressure. For the 48 samples suctioned successfully, the relationship between the weight of the fruit and the minimum negative pressure is shown in Fig. 10. The required negative pressure increased with weight: the heavier the fruit, the higher the negative pressure needed. The slope with the suction device oriented at 80 from the vertical was more than twice the slope at the vertical, indicating that in this device, the suction force in the approach position should be more than double that in the vertical position. This implies that the device should be able to transport a fruit in the vertical orientation if it is successful in the approach position.

3.4.
0

Picking-up performance Execution time

3.4.1.

-20

-40 -40

-20

20

40

Inclination from front view,


Fig. 9 e Inclinations of suctioned fruit from front and side views: n [ 30; B, Natsuakari; -, Deco rouge; , Benihoppe.

The mean execution time required for a picking-up operation was 8.9 s; this included 2.5 s for approach, 0.9 s for suction, 3.9 s for transportation, and 1.6 s for placement and return to the initial position. This study reveals a mechanisation process for picking-up, transferring and release of the strawberries by suction airow, and our results suggest a greater potential for mechanical packing if control of fruit posture is added after suctioning. Manual packing of two-layer boxes is estimated to take 8.3 s per fruit in Japan (Konya & Omori, 2010). Our result of 8.9 s is a little slower than for manual packing, and the execution time does not include alignment of the fruit in the box. A faster speed would be required for practical application of

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

355

5.0 4.5

100.0

Negative Pressure, kPa

3.5 3.0 2.5 2.0

Success rate, %

4.0

90.0

80.0
b

70.0

60.0
1.5 1.0 0.5 0.0 0 10 20 30 40 50 60

50.0 Natsuakari Deco rouge Tochiotome Beni-hoppe

Weight, g
Fig. 10 e Relationship between fruit weight and minimum negative pressure inside suction device: cultivar Benihoppe; n [ 48; B, suction device of 80 to vertical position, y [ 0.0887x, R2 [ 0.8689; , suction device of vertical position, y [ 0.0406x, R2 [ 0.8403.

Fig. 11 e Success rates for each cultivar in picking-up test: -, 16-mm xed height of suction device; , 19-mm xed height of suction device; letters at the right top of bars indicate that success rates with the same letters are not signicantly different at the 5% level according to the c2 test.

this technique. Considering that the most time-consuming process was the transportation, there is likely to be room for improvement by speeding up the motions of the x-, y-, and z-axis sliders of the manipulator, and by searching for an optimal path for transportation.

3.4.2.

Success rate of picking-up

In the picking-up performance test, fruits were not dropped during transportation even when the suction device tilted from the approach posture (80 from the vertical) to the vertical. The success rate for each cultivar is shown in Fig. 11. It was found that a success rate of more than 95% could be obtained in all cultivars if the suction device approached at the height of 16 mm. The high success rate noted for Benihoppe (100%) supports the results of the functional tests on effective space, and demonstrates that the suction device can be positioned within the effective space. Considering the effect of the approach height, the success rate decreased to 71.9% for Deco rouge: a signicant difference was obtained when the approach height of the suction device was 19 mm. This is presumably because the suction device approached from a position higher than the height of effective space: as Deco rouge has a long-tapered shape, a lower approach height would be more suitable. On the other hand, Natsuakari showed a continuously high success rate. Since Natsuakari has a rotund shape and its diameter is bigger, it appears that the height of effective space located is higher in this cultivar than for Deco rouge. It was observed that the cause of picking-up failure was roughness of the calyx side rather than fruit orientation error. The suction device was able to compensate for fruit orientation errors caused by the machine vision algorithm, and

successfully pick up most of the fruit. When the calyx side was dented or not uniformly curved, sufcient negative pressure could not be obtained, resulting in picking-up failure. It was also conrmed that the fruit itself moved or sometimes jumped towards the tip of the suction device due to the suction airow. This movement appears to suppress the occurrence of bruises or abrasions, and consequently no damage to the pericarp was observed immediately after the picking-up test. Moreover, no bruises were observed on the pericarp of the calyx side one week after keeping in cold storage. A comparison with non-suctioned fruits kept under the same conditions showed there to be no difference in degree of damage, although the colouring had progressed. Our study claried the importance of control of the suction device with a height of effective space of close to or slightly greater than half the fruit diameter. However, since there are individual differences in fruit diameter, an additional sensor to detect the fruit diameter or the fruit height, which has a certain relationship with the fruit diameter, would improve air picking-up performance by dealing with differences in fruit shape such as rotund, tapered and long-tapered types. The next direction of this study will be mechanical singlelayer packing onto a soft sheet with hollows, in which fruit alignment in the package will be an important technique. Further studies should focus on detecting the inclination of the strawberry fruit during transportation and on techniques for correcting the fruit posture.

4.

Conclusion

This study demonstrated the feasibility of gentle handling of strawberry fruit by a suction device. The developed picking-up equipment comprised a 4 DOF manipulator, a suction device, a machine vision system, a belt conveyor, and a control unit.

356

b i o s y s t e m s e n g i n e e r i n g 1 0 9 ( 2 0 1 1 ) 3 4 8 e3 5 6

The suction device had a tapered tube with an inner diameter of 25 mm, and generated a suction airow of approximately 45 l min1. The fruit placed on the conveyor was moved under the camera, and the machine vision system assessed the fruits orientation. The suction device was advanced towards the fruit along the line of the fruits orientation. A picking-up method is proposed in this study in which the fruit itself is moved towards the tip of the suction device by suctioning airow to prevent bruises or abrasions on the pericarp from the device pushing on the fruit. The effective space for suctioning was analysed. For the cultivar Beni-hoppe, the smaller the fruit, the larger was the effective space. It was found that the optimal height was around or slightly greater than half the fruit diameter; however, the permissible distance in the transverse direction was small. The inclinations of the suctioned fruit viewed from the front and side varied considerably; one reason seemed to be measurement error in fruit orientation by the machine vision. This indicated that the proposed picking-up method had some difculty in holding the fruit in a constant posture. In the approach position (80 from the vertical), the suction device required a suction force of more than twice that required in the vertical position. In picking-up performance tests for Natsuakari, Deco rouge, Tochiotome, and Beni-hoppe, when the approach height of the suction device was xed at 16 mm, our picking-up equipment achieved a picking-up success rate of more than 95% without dropping the fruit during transportation. The success rate decreased to 71.9% for the long-tapered Deco rouge, when the approach height of the suction device was 19 mm. No damage was observed to the tested fruits. The execution time for picking-up a fruit and placing it into a tray was 8.9 s, and the most time-consuming process was the transportation.

Acknowledgements
The authors gratefully thank Mr. Takanobu Yano for providing test samples of Natsuakari and Deco rouge, and for fruitful advice on the characteristics of Japanese cultivars.

references

Bato, P. M., Nagata, M., Cao, Q., Hiyoshi, K., & Kitahara, T. (2000). Study on sorting system for strawberry using machine vision (Part 2). Journal of the Japanese Society of Agricultural Machinery, 62(2), 101e110. Imou, K., Kaizu, Y., Morita, M., & Yokoyama, S. (2006). Threedimensional shape measurement of strawberries by volume

intersection method. Transactions of the ASABE, 49(2), 449e456. Ishii, T., Toita, H., Kondo, N., & Tahara, N. (2003). Deciduous fruit grading robots (Part 1). Journal of the Japanese Society of Agricultural Machinery, 65(6), 163e172. Kashiwazaki, M., Nagasue, T., Soutome, H., Nakajima, M., & Omori, S. (2007). Non-destructive measurement of strawberry fruit rmness (Part 1). Journal of the Japanese Society of Agricultural Machinery, 69(6), 49e56. Kitazawa, H., Ishikawa, Y., Nakamura, N., & Shiina, T. (2008). New type of packaging for preventing damage to strawberry fruits during transport. Food Preservation Science, 34(1), 19e23. Konya, T., & Omori, S. (2010). Development of packing equipment for strawberries. Journal of the Japanese Society of Agricultural Machinery, 72(1), 80e85. Konya, T., Omori, S., & Hayashi, S. (2007). Study on quality presentation technology of strawberries (Part 1). Journal of the Japanese Society of Agricultural Machinery, 69(3), 67e74. Liming, X., & Yanchao, Z. (2010). Automated strawberry grading system based on image processing. Computers and Electronics in Agriculture, 71(Suppl. 1), 32e39. MAFF. (2007). Vegetables. The 84th Statistical Yearbook of Ministry of Agriculture, Forestry and Fisheries, Tokyo, Japan, pp. 178. Monta, M., Kondo, N., Ting, K. C., Giacomelli, G. A., Mears, D. R., Kim, Y., et al. (1998). Harvesting end-effector for inverted single truss tomato production systems. Journal of the Japanese Society of Agricultural Machinery, 60(6), 97e104. Nagata, M., Bato, P. M., Mitarai, M., Cao, Q., & Kitahara, T. (2000). Study on sorting system for strawberry using machine vision (Part 1). Journal of the Japanese Society of Agricultural Machinery, 62(1), 100e110. Nagata, M., Kinoshita, O., Asano, K., Cao, Q., & Hiyoshi, K. (1996). Studies on automatic sorting system for strawberry (Part 2). Journal of the Japanese Society of Agricultural Machinery, 58(6), 61e67. Nagata, M., Kinoshita, O., Asano, K., Cao, Q., & Hiyoshi, K. (1997). Studies on automatic sorting system for strawberry (Part 3). Journal of the Japanese Society of Agricultural Machinery, 59(1), 43e48. Reed, J. N., & Tillett, R. D. (1994). Initial experiments in robotic mushroom harvesting. Mechatronics, 4(3), 265e279. Subrata, I. D. M., Fujiura, T., Yamada, H., Ishihara, N., Yuzawa, T., & Hida, M. (1998). Cherry tomato harvesting robot using 3-D vision sensor (Part 2). Journal of the Japanese Society of Agricultural Machinery, 60(1). 59e58. Tallada, J. G., Nagata, M., & Kobayashi, T. (2006). Non-destructive estimation of rmness of strawberries (Fragaria ananassa Duch.) using NIR hyperspectral imaging. Environment Control in Biology, 44(4), 245e255. Tedford, J. D. (1990). Developments in robot grippers for soft fruit packing in New Zealand. Robotica, 8, 279e283. Yamada, H., Tanaka, N., & Takada, S. (2009). Practical application of non-destructive analyzer of strawberry quality. Journal of the Illumination Engineering Institute of Japan, 93(5), 273e277.

You might also like