Commercial buildings are among the largest energy consumers in the built environment, yet most still rely on either manual thermostat adjustments — responsive to occupant needs but prone to energy waste — or static, schedule-based setpoint management with predefined temperature profiles that offer little adaptation to actual occupancy or individual comfort needs. Achieving energy efficiency, occupant comfort, and flexible operation of building energy systems therefore requires a design approach that strikes the right balance between automated setpoint management and occupant intervention, rooted in human interaction, where setpoint management becomes the primary interface through which occupants and their environment engage in a collaborative, rather than commanding, relationship. Existing office buildings, in particular, warrant attention due to their inefficiencies and dynamic operational characteristics. Addressing this gap calls for frameworks that close the loop between occupant behavior and system response during building operation, especially as advancements in the Internet of Things and digital interaction services expand the technical possibilities and economic viability. This work presents an exploratory study towards prototyping a human-in-the-loop thermostat control framework, demonstrated through its deployment in an existing office building. The study pursues two aims: (i) developing a minimal-feedback and interactive approach that collects occupant input, builds collective comfort profiles, and accommodates diverse user behaviors through adaptive interface design; and (ii) establishing a responsive control strategy that integrates occupancy detection and user-defined rules to proactively adjust temperature setpoints, balancing responsiveness with energy efficiency through setpoint management as the central coordination mechanism. The field measurements indicate an overall energy saving of 24 % compared to traditional fixed schedules, while the final occupant survey shows that 67 % of participants rated the system’s responsiveness and 55 % evaluated its learnability as high or very high. These results confirm that human-in-the-loop control can effectively balance energy efficiency with user satisfaction, offering a scalable path toward more responsive and sustainable building operation.
Karjou et al. (Mon,) studied this question.