Increasing energy demand, decarbonization commitments, and growing expectations for thermal comfort are driving the need for more adaptive and efficient climate control in residential buildings. This review synthesizes contemporary intelligent HVAC control strategies, including model-predictive control (MPC), deep reinforcement learning (DRL), data-driven forecasting, and hybrid approaches. Following PRISMA guidelines, a set of studies published between 2010 and 2025 was systematically screened and analyzed to identify the dominant methodological trends, data requirements, implementation architectures, and evaluation practices reported in the literature. This review highlights how these methods differ in modeling assumptions, computational complexity, robustness to uncertainty, and suitability for residential environments characterized by stochastic occupancy and heterogeneous building stock. In addition, we examine enabling technologies such as sensing infrastructures, pricing signals, and embedded computation, as well as barriers to real-world deployment, including data availability, interpretability, and integration with existing building systems. The findings provide a consolidated framework for understanding the capabilities and limitations of intelligent HVAC control and outline research gaps that remain for achieving scalable, user-centered, and energy-efficient operation in residential buildings.
Felez et al. (Wed,) studied this question.