本文へジャンプ

成果報告書詳細
管理番号20190000000294
タイトル*平成30年度中間年報 IoT推進のための横断技術開発プロジェクト/高速ビジョンセンサネットワークによる実時間IoTシステムと応用技術開発
公開日2019/6/6
報告書年度2018 - 2018
委託先名国立大学法人東京大学 日本電気株式会社 オムロン株式会社 株式会社エクスビジョン
プロジェクト番号P16007
部署名IoT推進部
和文要約
英文要約Title: Project to develop cross-sectoral technologies for IoT promotion / Real-time IoT System Using High-speed Vision Sensor Network and Development of Its Application Technology (FY2017-FY2020) FY2018 Annual Report

A. The proposition of a sensor network structure and the whole system architecture is our target for the real-time IoT system in the real world. This year we worked on inspecting the frame synchronization of a multiple cameras system and the precision of synchronizing a network with various sensors, both using PTP (Precision Time Protocol). Also several prior technologies other than PTP that are applicable to the structure of network architecture were studied.

B. The goal is to deploy High-Speed-Vision Software Development Kit (HSV SDK) as a standard development platform in a wide range of IA(industry automation) field. During FY2018, as the next edition of the HSV SDK, we developed a High-Speed-Vision Mount Camera Kit (HSV-MC1) with size 1/10 compact camera body. In HSV-MC1, camera becomes independent from the main body to make easy to incorporate into existing IA robot system. In parallel with developing HSV-MC1, three solution core modules have developed.

C. Realization of high-speed recognition and sorting of granular objects (1cm diameter, 2 shapes, 6 colors and 36 engraved characters) for real-time IoT system is our research target, that delivers new values for manufacturing. To discriminate multiple high-speed moving granular objects with multiple attributes, we did following things and achieved the goal (1,000fps tracking, 150~fps recognition, counting error under 5% and recognition accuracy above 90%).

D. High-speed sensor fusion technology enables the high-speed and robust picking application. In order to realize this technology, this year we worked on the following items:
1) Using the robot hand with multi-modal sensors created last year, we defined an abstracted state quantity and confirmed that high-speed picking is actually possible.
2) We created a FA system that imitates a real application and incorporated 1). Next year, we will verify the effectiveness of our research using this system.

E. The research and development of the man machine cooperation system is aimed that the super-precise work done by a skilled worker becomes possible by even an unskilled worker, using the high speed and high precision operational support based on the results of the study item A. This year we achieved the super-precise work (peg-in-hole using a 50μm peg and a 70μm hole) by the man machine cooperation system, and we developed more high-speed and wide stroke stage system to extend a working area.

F. In order to maximize the achievements of this project, we have operated an organization “WINDS Network (Network for World Initiative of Novel Devices and Systems)”. The organization supports commercialization of efforts from information sharing related to high-speed image processing. This year we performed three seminars and an explanatory meeting. Also we constructed a data base towards the next step of WINDS Network activity. The number of WINDS Network’s member is 189 organizations as of march 2019.
ダウンロード成果報告書データベース(ユーザ登録必須)から、ダウンロードしてください。

▲トップに戻る