Research database

Despite worldwide textile production continuing to grow, automation in this sector remains difficult due to its dynamic nature that includes constantly changing fabrics, patterns, and many other factors all related to the evolving fashion market. These complexities have led to a heavy reliance on human labor and contributed to offshoring outside Europe driven by cost pressures and the perception that technological innovations would require major process changes. This project aims at introducing AI in the textile industry complementing human activities with intelligent support systems that ease manual workload without the need for costly infrastructures. We will focus on wearable technologies (e.g. glasses, armbands) to collect exo- and ego-centric multimodal data (vision, audio, language, muscle activity signals) and study the actions of human workers while manipulating pliable materials. The data will be elaborated with modern AI-based foundation models to segment and track deformable objects and recognize actions. Detect actionable contact points can be an intermediate step that connects object segmention to action recognition. The long-term goal is to have models able to reason about human activities and elaborate augmented reality tools as well as collaborative robotic agents that understand human intentions when manipulating complex deformable objects to support textile manufacturing.

• Tatiana Tommasi (Principal Investigator)
• Raffaello Camoriano (Component of the research team)
• Carlo Masone (Component of the research team)