Developing an AI based Chatbot for Machine Operator Assistance
Mursaleen, Muhammad (2025)
2025
Automaatiotekniikan DI-ohjelma - Master's Programme in Automation Engineering
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2025-07-25
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202507257790
https://urn.fi/URN:NBN:fi:tuni-202507257790
Tiivistelmä
This thesis looks at the growing complexity of industrial machines and how it makes life more challenging for machine operators. Operators need fast and accurate information, whether it's about how something works, what setting to use, or even general company guidelines. To address this, developed and tested an AI-powered chatbot aimed at simplifying everyday tasks for operators. The chatbot is built using a Retrieval-Augmented Generation (RAG) setup, which means it can pull information with right context. This helps it find a balance between being quick and precise and also keeping transparency in mind.
During testing, the chatbot showed good results. In about 80\% of the cases, it was able to find and return information that actually matched what the user needed, even though working with data pulled from PDF manuals was not always easy. A big part of its success comes from how well the AI models handled the context they were given. When the chatbot found the right background information, both cloud-based and local AI models could create answers that were clear and useful. Out of all the models, GPT-4o-mini stood out by answering questions fast and with a high level of accuracy.
But there were some bumps along the way. Extracting clean, usable data from manuals turned out to be tricky. Sometimes, AI models struggled to make sense of the information, especially when running only on local machines with limited hardware. These issues meant that some queries didn’t get perfect answers, especially when the context was messy or incomplete.
Looking forward, there are some clear next steps. Future work should aim to improve how data is cleaned and organized before it goes into the chatbot, test different embedding techniques, and involve more real users to get a better idea of what works and what doesn’t. Security is another important area that could be improved, especially as these systems become more common in the industry.
In summary, this AI chatbot shows real promise in making machine operators' jobs easier. By giving them fast and reliable access to the information they need, it has the potential to boost efficiency and reduce unnecessary downtime on the factory floor.
During testing, the chatbot showed good results. In about 80\% of the cases, it was able to find and return information that actually matched what the user needed, even though working with data pulled from PDF manuals was not always easy. A big part of its success comes from how well the AI models handled the context they were given. When the chatbot found the right background information, both cloud-based and local AI models could create answers that were clear and useful. Out of all the models, GPT-4o-mini stood out by answering questions fast and with a high level of accuracy.
But there were some bumps along the way. Extracting clean, usable data from manuals turned out to be tricky. Sometimes, AI models struggled to make sense of the information, especially when running only on local machines with limited hardware. These issues meant that some queries didn’t get perfect answers, especially when the context was messy or incomplete.
Looking forward, there are some clear next steps. Future work should aim to improve how data is cleaned and organized before it goes into the chatbot, test different embedding techniques, and involve more real users to get a better idea of what works and what doesn’t. Security is another important area that could be improved, especially as these systems become more common in the industry.
In summary, this AI chatbot shows real promise in making machine operators' jobs easier. By giving them fast and reliable access to the information they need, it has the potential to boost efficiency and reduce unnecessary downtime on the factory floor.