Residential heating, ventilation, and air conditioning (HVAC) systems constitute a significant proportion of the building's energy usage and require optimisation of energy management. In this context, occupancy HVAC control is a promising option with energy savings of 20-50% in the home. However, occupancy sensing technology suffers from long retrieval times, privacy issues and low comfort. Additionally, there is an increasing need for even more advanced technologies that can help regulate indoor air quality in addition to energy control.
To meet these expectations, scientists have recently changed to intelligent control methods such as quantum reinforcement learning (QRL) based on the principles of quantum computing. Such an approach can particularly accelerate machine learning processes and handle the complexity of real building dynamics.
In the new breakthrough, a group of Korean researchers led by Sankeum Lee, assistant professor of computer engineering at Hanbat National University, has presented the first demonstration of ongoing quantum-enhanced reinforcement learning for residential HVAC and residential power management. Their innovative findings were made available online on June 16, 2025 and published in Volume 21 of Journal Energy and AI on September 1, 2025.
Dr. Lee emphasizes the novelty of their work. “Unlike traditional reinforcement learning technology, QRL leverages quantum computing principles to efficiently process high-dimensional states and action spaces, enabling more accurate HVAC control in multi-zone homes. Our framework uses operational data, including consumption electronic control data, external temperature fluctuations, and integrates real-time occupancy detection using operational data.
Additionally, the proposed technology integrates features such as multi-zone cooling, controls the temperature and clustering of individual zones in the building, group similar data points, and regulates cooling. In this way, a single controller collaboratively optimizes comfort, energy costs and carbon signals in real time.
Researchers conducted simulations based on real-world data from 26 residential households over three months. They found that QRL HVAC control significantly outperforms the deep deterministic policy gradient method and proximal policy optimization algorithms, whilst achieving a 63% and 62.5% reduction in power costs, while maintaining thermal comfort, while achieving a 63% and 62.4% reduction in power costs, respectively.
There are many more advantages to the current approach. It is retrofit-friendly and works with standard temperatures, occupancy rates, CO₂ sensors, general HVAC equipment and thermostats. It is also robust against uncertainty and easily handles noisy predictions about weather, occupancy and device constraints. Additionally, there is a generalizable framework that can be extended from apartments to small buildings and microgrids.
Dr. Lee talks about the potential applications of their innovation. “It can be used with smart thermostats and autonomous home energy management systems that co-optimize comfort, billing and emissions without manual tuning, rooftop solar and household battery scheduling. Our framework also helps with utility demand response and time usage programs with automatic control.”
QRL-based HVAC controls are especially applicable on a community or campus scale via grid interactively efficient buildings and virtual power plants (VPPs). Here, millions of homes can be tailored as VPPs to stabilize a grid that is rich in renewable energy. It also allows you to ensure personalized indoor environmental quality within your carbon budget and integrate advanced intelligent control options.
As hardware matures over the next few years, quantum accelerated policy searches could quickly facilitate training of complex multi-energy systems such as HVAC, electric vehicles, and energy storage systems. In the long run, this work is expected to guide the path to a standardized, secure controller that can be certified and deployed on a wide range of scales!
/Public release. This material of the Organization of Origin/Author is a point-in-time nature and may be edited for clarity, style and length. Mirage.news does not take any institutional position or aspect, and all views, positions and conclusions expressed here are the views of the authors alone.
