Modeling and Control Beyond Static Boundaries: Integrating human-building interaction, adaptive comfort, behavioral intelligence, and grid-responsive control.
About the Workshop
The energy transition creates a critical opportunity to rethink how buildings model, predict, and respond to occupant behavior. From personalized thermal comfort control and human-in-the-loop systems to data-driven behavior modeling and adaptive automation, emerging technologies enable buildings to become truly occupant-centric energy systems.
Occupant-aware approaches are essential for:
- Providing thermal, visual, and acoustic comfort while reducing energy waste
- Enabling demand response and renewable energy integration
- Supporting grid flexibility through intelligent load coordination
- Addressing equity in comfort provision across diverse populations
- Enhancing resilience during extreme weather events and grid disruptions
Despite rapid advances in reinforcement learning, IoT-enabled automation, and smart building control, occupant behavior modeling remains fragmented across disciplines. OccuSys brings together researchers and practitioners working at the intersection of human behavior, building systems, and energy management.
The workshop aims to:
- Share frameworks, datasets, and best practices for occupant-centric modeling and control
- Identify research gaps and future directions in adaptive comfort and human-aware energy systems
- Foster interdisciplinary collaboration between academia and industry
Program
Keynote Speakers
Prof. Dolaana Khovalyg
EPFL – Laboratory of Integrated Comfort Engineering (ICE)
Closing the Loop on the Human Body: Sensing, Modeling, and Control for Occupant-Centric Buildings
Abstract:
Occupant-centric building research has advanced rapidly in areas such as occupancy detection, preference learning, reinforcement learning-based HVAC control, and personal comfort systems, yet these advances remain largely disconnected.
The human body, one of the most informative signal sources in a building, is still treated as a black box inferred indirectly through surveys or ambient sensors.
Closing this loop requires coordinated progress across three tightly coupled layers: scalable physiological sensing, personalized models that translate sparse biomarkers into actionable estimates of thermal state, and adaptive controllers that act on these estimates to balance energy, comfort, and well-being.
This keynote presents an integrated view of how these layers can work together, drawing on research at the Laboratory of Integrated Comfort Engineering (ICE) at EPFL.
Prof. Misha Pavel
Northeastern University
Human-System Identification and Control: Lessons from Healthcare to Grid-Interactive Buildings
Abstract:
Smart health systems provide a valuable model for a central challenge in smart buildings: how to identify, predict, and improve human-system behavior in everyday environments. In healthcare, system identification, control theory, and unobtrusive sensing have been used to model patient behaviors, detect meaningful changes in routines, and design personalized interventions.
This presentation examines human-system identification and control across smart health and the smart built environment. We explore control-theoretic models of human behavior used to predict and improve patient health behaviors, and we consider how similar ideas can support occupant modeling in buildings. We also survey the broader smart health landscape, emphasizing shared challenges such as noisy field data, individual differences, delayed feedback, privacy, and interventions that must work within daily life.
We reflect on applying these ideas to a longitudinal study of occupant thermostat use and other human-building behaviors in homes. Finally, we contrast the strengths and pitfalls of these distinct modeling philosophies, providing a definitive roadmap for modeling human behavior to improve well-being of people and the built environment.
Topics of Interest
We invite original research, position papers, and work-in-progress submissions on all aspects of occupant behavior modeling and control for energy management. Topics include, but are not limited to:
- Modeling Approaches: Data-driven vs. physics-based vs. hybrid models; Personal comfort models and preference learning; Occupancy detection and prediction methods; etc.
- Control and Optimization: Occupant-centric controls; Adaptive and personalized comfort control; Model predictive control with behavior forecasting; Privacy-preserving controls; etc.
- Beyond Static Comfort Boundaries: Adaptive thermal comfort models in practice; Context-dependent comfort; Thermal comfort equity and individual differences; etc.
- Data and Benchmarking: Longitudinal studies and real-world deployment; Wearable sensors and physiological monitoring; Privacy, ethics, and user acceptance; etc.
- Integration with Energy Systems: Demand response enabled by occupant behavior prediction; Thermal energy storage and occupant flexibility; Building-to-grid services considering occupant comfort; etc.
- Theory and Methods: Interpretability and explainability in ML-based models; Validation and verification methodologies; Interdisciplinary approaches; etc.
Submission
Submissions must be unpublished and not under review for any other venue. Papers must be at most 4 pages (single-spaced, US Letter 8.5” × 11”), including figures, tables, and appendices. The format must follow the official ACM proceedings template (sigconf format) (LaTeX preferred or Word) and comply with ACM formatting requirements (9-pt font). Authors must anonymize their manuscripts by enabling the anonymous option and using the anonsuppress section where appropriate. Papers that do not comply with size, formatting, or anonymization requirements will not be reviewed.
All papers must be in Portable Document Format (PDF) and submitted via
Registration is handled through
buildsys.acm.org/2026
.
Important Dates (AoE)
- Submission: April 17, 2026
- Notification: April 24, 2026
- Camera-ready: May 5, 2026
Organizing Committee
- Prof. Zoltan Nagy (TU Eindhoven)
- Prof. Michael Kane (Northeastern University)
- Dr. Martin Mosteiro Romero (TU Delft)
- Dr. Wei Luo (TU Eindhoven)
- Maharshi Pathak (Northeastern University)
- Ava Mohammadi (TU Eindhoven)
- Mahnaz Vahdat (Northeastern University)
Contact
For questions, please contact:
z.nagy@tue.nl