Research Lines & Projects
Applied Artificial Intelligence, Swarm Robotics, Reinforcement Learning, and Building Decarbonization.
My research at Tecnalia and UPV/EHU tackles foundational and applied challenges in intelligent decarbonization and construction automation. Our work spans multi-agent swarm robotics, deep reinforcement learning, and full-scale empirical validation.
SITEBOT: Swarm Robotic Intelligence for Timber Building On-site Technologies
SITEBOT targets safer, faster, and greener on-site building assembly by advancing autonomous robot collectives for timber-based construction. Funded under the European Union’s Horizon Europe Research and Innovation Programme (EIC Pathfinder Challenges, Grant Agreement No. 101307035), the project develops reconfigurable multi-robot collectives capable of cooperative manipulation, adaptive perception, and decentralized decision-making in unstructured building construction environments.
Key Contributions & Highlights:
- Funded by the European Union’s Horizon Europe EIC Pathfinder Challenges (Grant No. 101307035).
- Multi-agent coordination and collective intelligence for sustainable timber assembly.
- Cross-disciplinary European consortium across advanced robotics, automation, and sustainable construction (2026–2030).
Reinforcement Learning for Building Energy Management
Modern buildings feature significant thermal inertia and non-linear dynamics. We apply state-of-the-art Deep Reinforcement Learning (DRL) algorithms (such as DQN, PPO, DDPG) combined with generalized reward functions to train intelligent energy operators capable of adapting to shifting weather patterns, dynamic electricity tariffs, and occupant preferences.
Key Contributions & Highlights:
- Novel training workbenches using transient building energy models.
- Reward formulations balancing indoor thermal comfort and grid cost.
- Validation against standard baseline rule-based controllers.
Model Predictive Control (MPC) & Evolutionary Optimization
Integration of intermittent renewable generation and price-responsive demand requires anticipating thermal loads and energy market fluctuations. Our research explores evolutionary algorithm-based MPC formulations for hybrid electrical-thermal microgrids, handling complex multi-objective optimization under uncertainty.
Key Contributions & Highlights:
- Short-term predictive scheduling for thermal storage and heat pumps.
- Integration with day-ahead and intra-day electricity market prices.
- Multi-zone building co-simulation.
Experimental Validation at KUBIK Test Facility
KUBIK is Tecnalia's unique experimental building infrastructure dedicated to R&D in energy efficiency and building envelope technologies. We implement and empirically validate our AI-driven control architectures in real-world scenarios, bridging the gap between numerical simulation and physical deployment.
Key Contributions & Highlights:
- Real-world test campaigns under varying realistic boundary conditions.
- Continuous sensor telemetry, energy metering, and SCADA integration.
- Comprehensive case study featured in international publications and book chapters.
Advanced Control of Ground Source Heat Pumps (GSHP)
Ground source heat pumps represent one of the most efficient pathways for heating decarbonization, yet their high thermal inertia and soil regeneration dynamics complicate conventional control. We research data-driven and adaptive supervisory controls to maximize seasonal coefficient of performance (SCOP).
Key Contributions & Highlights:
- Comprehensive state-of-the-art review published in Renewable and Sustainable Energy Reviews.
- Long-term borehole thermal stability and energy efficiency analysis.