At the CoDIT 2025 conference Tomasz Kołcon form Łukasiewicz-PIAP Institute, and Krystian Goławski from VIGO Photonics presented the research and results from Polish pilot site of AI-PRISM project.
In the AI-PRISM project, we explore how Artificial Intelligence (AI) and Human–Robot Collaboration (HRC) can enhance small-scale, high-precision manufacturing processes. At VIGO Photonics, short production runs make full automation impractical, leaving many delicate tasks (such as semiconductor lens shaping) performed manually. These processes are repetitive, demanding, and prone to human error. Our goal is not to replace operators but to assist them through semi-automated systems designed to improve accuracy, reduce strain, and make production more sustainable.
To address these challenges, we developed an advanced assembly station featuring motorized XY tables with sub-micron precision, a multi-camera setup for improved visibility, and AI modules for automatic chip positioning and defect detection. This setup reduces the operator’s need for constant manual alignment and visual inspection. A three-stage engagement programme involving task analysis, eye-tracking, and interactive workshops ensured that the system was built with user needs at its core. By directly involving operators, we identified challenges such as microscope calibration and glue application, leading to practical design improvements like robotic positioning aids and automated error detection.
Testing the new AI-assisted station revealed significant benefits: lower mental strain, improved usability, and faster learning curves for novice operators, while maintaining physical workload at a safe level. Physiological and self-assessment data confirmed reduced cognitive load and enhanced user experience. These findings underscore the potential of AI-driven solutions to not only increase manufacturing efficiency but also improve the well-being and engagement of workers. Looking forward, AI-PRISM will continue exploring long-term user acceptance, the role of AI trust and explainability, and how these innovations can be adapted to other high-precision manufacturing contexts.