In contemporary industrial environments increasingly shaped by automation and human-robot collaboration, the objective and continuous assessment of operator cognitive workload and stress represents a critical challenge for both safety and productivity. This thesis contributes to the development of a multimodal wearable system for the near real-time monitoring of cognitive and emotional states in logistics operators, with a specific focus on the computer vision module and oculometric analysis. The hardware platform consists of a sensorized safety helmet integrating a multimodal acquisition system comprising electroencephalography (EEG), electrodermal activity (EDA), photoplethysmography (PPG), and a dual-camera computer vision module. A Raspberry Pi 5 mounted on the helmet manages real-time video acquisition, oculometric feature extraction, and data storage, ensuring the operator's full freedom of movement.