Proceedings of the
9th International Conference of Asian Society for Precision Engineering and Nanotechnology (ASPEN2022)
15 – 18 November 2022, Singapore
doi:10.3850/978-981-18-6021-8_OR-05-0222
Automated knowledge extraction for smart cognitive Machine tool HMI
1Department of System Engineering, Kyung Hee University, 1732 Deokyeong-daero, Giheung-gu, Yongin-si, Gyeonggi-do, Korea
ABSTRACT
The machine tool market is formed with general-purpose machines and dedicated machines developed by various vendors. Therefore, it is common to use a mix of machine tools from various makers in the single shopfloor. However, diverse manufacturers have their own interfaces, and there are a lot of different versions. Their instructions made of large-size files or heavy paper handbooks are provided to the users, but portability, storage convenience, and visibility are vulnerable. Machine tool users need separate long-term training or trial-and-errors to use the machine without helps. Meanwhile, it is becoming difficult to find technically trained manpower due to population changes in the manufacturing-advanced countries. Automated factories using machine tools is always suffering a shortage of expertized labors, and training courses for new workers are not properly prepared. To address these situations, proposals and use of Digital Intelligent Assistant for factory automation are increasing. In this paper, we propose an automated knowledge extraction process consisted of three interpretation functions. The machine interpretation function uses a machine interface library to read machine status information from the machine. The user Interpretation feature uses the machine simulator image for optical character recognition to identify the user's intention and the operation currently being performed. The manual interpretation function utilizes an information retrieval algorithm that searches text information from large machine instructions. It also uses reinforcement learning to shorten long-term training and replace unnecessary trial-and-errors. New cognitive interface supported by automated knowledge extraction functions can easily interact with humans and provide appropriate answers in intuitive ways. The proposed system will be the basis for integrated cognitive human-machine-interface regardless of the manufacturer. Through this, the efficiency of limited manpower will be maximized in industrial sites where various interfaces are mixed, and human with factory will be realized.
Keywords: Integrated, HMI, Cognitive, Machine, Reinforcement Learning, Vision, NLP
1Department of System Engineering, Kyung Hee University, 1732 Deokyeong-daero, Giheung-gu, Yongin-si, Gyeonggi-do, Korea
ABSTRACT
The machine tool market is formed with general-purpose machines and dedicated machines developed by various vendors. Therefore, it is common to use a mix of machine tools from various makers in the single shopfloor. However, diverse manufacturers have their own interfaces, and there are a lot of different versions. Their instructions made of large-size files or heavy paper handbooks are provided to the users, but portability, storage convenience, and visibility are vulnerable. Machine tool users need separate long-term training or trial-and-errors to use the machine without helps. Meanwhile, it is becoming difficult to find technically trained manpower due to population changes in the manufacturing-advanced countries. Automated factories using machine tools is always suffering a shortage of expertized labors, and training courses for new workers are not properly prepared. To address these situations, proposals and use of Digital Intelligent Assistant for factory automation are increasing. In this paper, we propose an automated knowledge extraction process consisted of three interpretation functions. The machine interpretation function uses a machine interface library to read machine status information from the machine. The user Interpretation feature uses the machine simulator image for optical character recognition to identify the user's intention and the operation currently being performed. The manual interpretation function utilizes an information retrieval algorithm that searches text information from large machine instructions. It also uses reinforcement learning to shorten long-term training and replace unnecessary trial-and-errors. New cognitive interface supported by automated knowledge extraction functions can easily interact with humans and provide appropriate answers in intuitive ways. The proposed system will be the basis for integrated cognitive human-machine-interface regardless of the manufacturer. Through this, the efficiency of limited manpower will be maximized in industrial sites where various interfaces are mixed, and human with factory will be realized.
Keywords: Integrated, HMI, Cognitive, Machine, Reinforcement Learning, Vision, NLP
