Trust is essential for effective automation. In Human-Robot Interaction, trust becomes more complex due to the robot's physical presence alongside the operator. Current research relies on post-hoc trust measurements, which help optimize system design but don't allow for real-time trust adjustments during tasks. This project aims to: (i) identify trust indicators in industrial Human-Robot Collaboration, (ii) collect sensor data and train machine learning models for real-time trust prediction, and (iii) adapt robot behavior based on predicted trust levels to ensure efficient task performance.