Workers in advanced recycling manufacturing face a variety of risks. These include exposure to hazardous chemicals, dust, and biological agents, in addition to physical hazards associated with heavy machinery and repetitive movements. These risks can lead to respiratory problems, skin irritation, physical injuries, and chronic health issues. HARMONICA’s unique vision is to transform current cognitive and physical augmentation technologies into more supportive tools for operators within the contexts of hazardous environments such as advanced recycling manufacturing.

To do this, HARMONICA starts from a human centric perspective, enabled by digital twin modelling. This will enhance the safety and efficiency of human operators in dynamics and hazardous environments through a Psy-PhyTwin, which embeds cognitive and physical submodels of workers into an industrial digital twin. This is a paradigm shift from machine-centric to a human-inclusive ecosystem in which people and technology develop together. Context-aware perception enhanced by photonics provides precise insight into shop-floor dynamics. Semi-autonomous mobile manipulators (AMMs) will be developed to relieve workers from the more dangerous and fatiguing tasks, enabling them to focus on skilled, creative, and supervisory roles. A mixed-reality learning environment accelerates training and continuous upskilling for workers, managers, and supervisors, supporting coaching, workflow optimisation, and process safety. This will be based on project work that moves beyond efficiency-driven paradigms towards more human centric work spaces and tooling. The guiding principles within HARMONICA are participatory co-design, explainable and controllable technology, equitable access to training, lifelong learning, and alignment of digital innovation with environmental and societal sustainability.

MKLab has a key scientific and technical role in the HARMONICA project by leading WPs 4 and 5 on multimodal sensing and Perception, as well as tasks on quality assurance and risk management, real-time physiological sensing and teaming, object-centric scene perception, and the adaptive vision companion. The developed solutions cover multimodal sensing, human cognitive and physiological state assessment, human action and intention understanding, industrial scene perception, and explainable visual assistance for safer and more effective human–machine collaboration.