工作日誌登載系統

專利類型

發明

專利國別 (專利申請國家)

中華民國

專利申請案號

109104894

專利證號

I 767189

專利獲證名稱

工作日誌登載系統

專利所屬機關 (申請機關)

淡江大學學校財團法人淡江大學

獲證日期

2022/06/11

技術說明

本計畫為一年期的產學合作計畫，其合作的對象為程曦公司，是個專業從事智慧客服及銀行對話服務機器人的公司，近幾年也從事研發人工智慧在機器人服務之技術與應用。在研發的過程中，遭遇機器人的機型不同，硬體設備不同(四肢、移動方式)、內建服務功能不同、作業系統及軟體支援程度不同、以及支援之API與程式語言的不同等，使得服務機器人客製化的過程中遭遇編程設計繁瑣、程式過度仰賴硬體及系統支援、因機器人硬體而限制了工程師的開發空間、跨系統及跨機器人平台時(機器人硬體版本更新或品牌更換時)，軟體無法複製或重覆使用等問題，這些問題也造成機器人應用工程師面臨的軟體開發時間長、受硬體限制、開發效能低落、多版本不易維護等問題。 本產學案基於上述動機，擬協助程曦公司，進行三項重要的研發工作。首先，在第一項工作中，我們擬定義一套可適用於跨平台及多種機器人的虛擬語言，這樣的語言可表達大多數機器人可執行的指令與動作，並能表現平行及循序執行指令的關係，例如，機器人在服務銀行櫃台時，可邊說「請到五號櫃台」、身體邊轉向五號櫃台、而手也同時指向五號櫃台。其次，在第二項工作中，我們將在實體機器人中設計一個Adaptor，這個Adaptor有能把虛擬語言轉譯於成機器人能執行的語言並自動呼叫該機器人所支援之API，以便使機器人的各項軟硬體設備可實現虛擬語言所製定的動作與指令，對於這項工作，我們將以華碩機器人Zenbo為實證場域，在Zenbo機器人上開發Adaptor，以驗證虛擬語言可確實執行於實體機器人。最後，在第三項工作中，我們將在Zenbo機器人中，開發一套客製化的機器人服務，使其能在公司迎賓大廳中對員工進行「上班刷臉打卡」及對新客戶進行公司導覽解說之功能。 透過這個產學合作案，我們所研發的三項工作，包括(1)機器人虛擬語言；(2)在Zenbo機器人中開發Adaptor；以及(3)客製化的機器人「上班刷臉打卡」及「公司導覽解說」之功能，將可解決程曦公司，甚至是對服務機器人產業，解決其面臨多種機器人開發的痛點，並使機器人客服開發的軟體得以重覆使用及易於管理與維護，而對於機器人硬體更換所需之軟體跨平台之移植亦可較有系統的提供解決方案，進而帶動台灣在機器人服務的產業發展與技術領先，也間接有助於台灣物聯網與人工智慧應用於機器人之產業發展。 This project proposes a one-year industry-university cooperation plan. Its cooperation company is Chain-Sea company, a company specializing in intelligent customer service and bank Chat bot service. With the development of Internet of things and artificial intelligence technologies, a variety of robots also appear in the market and are gradually applied in the service industry such as shopping malls, reception, ordering food and so on. Because that different robots support different hardware, platforms, software, functions, API as well as programming languages, several challenges have been encountered when developing customization service using robot. These challenges include cumbersome programming design, programming highly dependent on hardware and system supports and difficulties for software reuse. These challenges also lead to the problems faced by robot application engineers, such as long software developing time, low developing efficiency, and difficult maintenance for robot service systems. The project aims to cooperate with research team of Chain-Sea company to design and implement three important tasks. First of all, we aim to define a cross-platform virtual language which can present most robots executable instructions and actions, and can represent parallel and sequential relationship between instructions or functions. For example, a robot in the bank counter can perform three actions: "say: please go to counter number five", “turn its body side to counter number five,” and “hands pointing to counter number five” at the same time. In the second task, we will design an Adaptor in the real robot. The Adaptor can translate instructions represented in virtual language to the instructions of robot supported language. The translated instructions can automatic call APIs supported by the robot, in order to perform the instructions of virtual language in real robot. To achieve this, we will implement the Adaptor in ASUS robot Zenbo for empirical field, to verify that the virtual language can carry out in physical robot. Finally, in the third task, we will develop a customized robot service in Zenbo robot, which can support the service of "take picture and punch in" for employees in the reception hall of the company and provide guidance and explanation for new customers. The execution of this project will help Cheng Xi company, even for robot service industry, improve the efficiency of developing, reusing as well as maintaining robot service software/systems. In addition, it also reduces the difficulty for robot hardware replacement required software cross-platform transplantation. The execution of this project can also promotes the development and leading technology of robot service industry as well as contribute to the industrial development of the Internet of things and artificial intelligence applied to robots in Taiwan.

備註

連絡單位 (專責單位/部門名稱)

研究發展處

連絡電話

26215656分機2561

網址