Automated kitchen assistant intelligent software

• Autor(a) principal: Rosa, Ana Cláudia Ferreira
• Data de Publicação: 2024
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Texto Completo: http://hdl.handle.net/10773/44659
• topic: Software development; Time series data; Statistical models; Neural networks; Demand forecasting; Internet of things; Ingredient inventory management

The rapid advancement and globalization of technology have transformed the food service industry, particularly through the emergence of automated kitchens that integrate robotics and smart technologies. These fully robotic systems enable the mass production of customized meals, addressing critical challenges such as ingredient management and food waste reduction. This dissertation presents the development of an intelligent software designed to optimize operations in automated kitchens. By leveraging AI and IoT, the software manages ingredient inventory, forecasts demand based on consumption patterns, and generates notifications to streamline supply processes. The development followed an agile methodology, starting with core functionalities for ingredient reception, purchasing and storage, and ultimately generating reports to track consumption and waste. The system supports three user roles: administrator, food handler, and store manager. In its final phase, a forecasting system was integrated to predict daily ingredient consumption. Two models were implemented, a statistical VAR model and an LSTM neural network, trained and evaluated using time series data. A rolling cross-validation technique with three folds was applied to refine model performance and hyperparameters. Despite the challenge of limited data, the LSTM model outperformed the VAR model, and its forecasts were integrated into the software to generate supply notifications. Usability tests involving five participants confirmed the system’s user-friendly design and essential features for managing an automated kitchen. However, the tests also identified some usability issues and bugs, revealing areas for further improvement to enhance user experience and operational efficiency. Overall, this research contributes to advancing inventory management in automated kitchens by providing a scalable, adaptable solution that meets the evolving needs of the food service industry.