AI-Driven Waste Classification and Smart Management in Poland’s Industry 4.0 Framework: CNN-Based Solutions for Sustainable Municipal Systems

The article is devoted to the integration of artificial intelligence (AI) into municipal waste management systems in the context of Industry 4.0, with a focus on challenges and opportunities for Poland. It analyzes key technologies such as the Internet of Things (IoT), smart containers, optimization of waste collection routes, and automated sorting, giving examples of implementations in European cities (Barcelona, Copenhagen, and Amsterdam). The article develops and evaluates a prototype of a convolutional neural network (CNN) based on MobileNetV2 for waste classification based on photographs from the Kaggle dataset (over 19 thousand images in 10 categories). The model, trained using transfer learning, data augmentation, and class balancing, achieved 93% accuracy on test data, with high precision, recall, and F1-score; the error matrix showed minimal confusion between similar classes (e.g., plastic and glass). The results confirm the potential of AI to improve sorting efficiency and reduce costs (up to 30%) and environmental emissions, contributing to the EU objectives of the circular economy and the waste hierarchy in Poland. The article suggests areas for further research: long-term evaluations, model adaptation, and ethical aspects, highlighting the role of AI in sustainable waste management.