2005 Fiscal Year Final Research Report Summary
Development of information accumulated and automatically supplied grading robot for multi agriculture products
| Project/Area Number |
16580216
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Agricultural information engineering
|
|---|
| Research Institution | S-I Seiko Company, Limited |
|---|
Principal Investigator |
NINOMIYA Kazunori S-I Seiko Company, Limited, SI SEIKO Co., an assistant section manager (20374150)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KAMATA Junzou SI SEIKO Co., 技術開発部, a subsection chief
KURITA Mitsutaka SI SEIKO Co., 技術開発部, chief (00451302)
RAJENDRA Kumar Peter エスアイ精工株式会社, 技術開発部, 主任
ARAKI Wakako SI SEIKO Co., chief
|
|---|
| Project Period (FY) |
2004 – 2005
|
| Keywords | Grading Robot / Trace-ability / Food safety and relief / Neural network / Image processing / Direct lighting |
|---|
| Research Abstract |
A series of grading robot for multi agricultural products is reported. At first, a grading robot is manufactured experimentally. The robot has a CCD camera that is on conveyer that moves objective agricultural product at random. Size and position on the conveyer, of objective product are detected by image processing captured by the CCD camera. After that detection, the agricultural product is moved by multi joint robot which uses suction pad properly. In the moving process, the agricultural product is captured full surface except suction pad face by another CCD camera, and inspected its grade about color, shape and defect. After then, it is stored to shipping box. The stored product's grading information is registered database, and utilized to search by using ID that is printed in its box. When another shape type is used, the suction and identify code for image processing are needed to change.
Secondly, as for grading method by image processing, effectiveness of direct lighting by use of PL filer for glossy fruit, is indicated and mass estimation for multiple agricultural products is experimented. And grade judging algorithm by use of data that is obtained by image processing A judging method is implemented by using neural network. The neural network whose inputs are evaluation values from image processing(shape, defect etc.), and whose output is grade-level, are constructed.
Finally, applications of grading robot for multi agricultural products are considered. By realizing information accumulated grading robot, it could be linked shipped agricultural product to various information. It leads effectiveness of food safety and relief Besides, by parallel working the grading robot, it is high efficient and can make space small in facility.
|
Research Products
(14 results)
