Research

Planetary stewardship refers to the intentional monitoring, modelling, and protection of Earth’s complex life-support systems. As human activity pushes planetary boundaries, AI is evolving beyond a commercial tool into a crucial socio-ecological infrastructure.

We envisage [ai] as a lense for stewardship operating across three interconnected dimensions:

Ecological AI: Advanced detection, modeling, and decision-support technologies for monitoring biodiversity, ocean conservation, resilient and sustainable coastal urban systems, and the energy systems of the future.

Physical AI: Next-generation autonomous devices and cognitive artifacts with embedded intelligence that bridge the physical and digital worlds.

Civic AI: Human-centered computing and ecological-context-aware systems developed through participatory co-design methods, citizen engagement, and governance models that integrate ethical and safety principles into AI development.

Our research is organised in four pillars. The first three map onto the AI dimensions and the fourth cuts across all of them.

---

Innovation in Digital Twins of Sustainable Energy Systems

Ecological AI · Activity A1 · M3–M36 · Building on AHEAD, Shift2DC, ALAMO

The Port of Funchal is at once an economic gateway and a large energy consumer within an isolated island system. We build digital twins — computational replicas fed by real-time data — at two interlocking scales: of the port, integrating energy, environmental, mobility and infrastructure data; and of the adjacent energy communities, modelling the port as a renewable producer and the surrounding urban area as consuming members of a collective self-consumption community.

The aim is to make it possible to assess scenarios before executing them — quay electrification, peak management, surplus sharing — and to demonstrate the viability of urban energy communities where physical constraints limit renewable installation.

Innovation in Health and Wellbeing for Resilience

Physical AI · Activity A2 · M3–M36 · Building on eGames Lab, AppCovidMadeira, MICROCLI-MAC, SANIFISH

Madeira is ageing, receives millions of visitors a year, and faces health risks specific to a subtropical island region. A reactive health system responds late to each of these; a data-informed one can anticipate them. We develop digital solutions that combine individual, environmental and social data to generate personalised recommendations, including predictive monitoring of epidemiological vectors using AI and environmental sensors. In parallel, we build and pilot adaptive serious games with wearables and biofeedback, anchored in regional culture, for stimulating and rehabilitating cognitive and motor skills in older adults.

Innovation in Interaction for Civic Preparedness

Civic AI · Activity A3 · M3–M36 · Building on eGames Lab, Sense and Sensibility in Interactivity

Civic preparedness today is not basic digital literacy: it is the capacity to recognise AI-generated content, to participate knowledgeably in debates about regulation, and to hold developers and deployers to account. Without it, the digital transition happens to communities rather than with them. We work in co-design with communities and domain experts to identify the problems that matter, build prototypes and interactive experiences that use AI, and evaluate them with real users. One strand turns digital twin visualisations — normally built for experts — into experiences the public can explore; another works on civic participation in public health and wellbeing.

Ethical and Sustainable Artificial Intelligence

Cross-cutting · Activity A4 · M3–M36

AI’s potential to build resilient communities is clear. So is its cost: data centres consuming water and energy at scale, models trained on sensitive material, systems whose decisions nobody can explain. We establish practices to identify and mitigate bias, ensure explainability and protect privacy, in line with the European AI Act. And we quantify the energy use and CO₂ emissions of training and inference across our own models, investigating reduction strategies — algorithmic optimisation and training management tied to renewable energy availability.

---