A Literature Review on the Application of Physical AI Robotics for Zero-Energy Building Performance Gap

Zero-energy buildings (ZEBs) exhibit energy performance gaps (EPGs) of +14-+30% between design predictions and actual operations.This study reviewed 27 publications (19 peer-reviewed articles and eight technical/industrial reports, 2022-2025) across three domains, physical AI robotics, physics-informed neural networks (PINNs), and building digital twins, to analyze the applicability of emerging physical AI technologies to ZEB performance gap closure.Existing reviews on physics-informed machine learning (PIML) focus on methodological taxonomy without addressing robotics or ZEB performance gaps, whereas ZEB digital twin studies rely on fixed Internet of Things (IoT) sensors without autonomous robotics and BIM-robot integration studies are limited to indoor environment monitoring without energy optimization linkages.This review reveals that no prior study has integrated these domains within a unified ZEB performance management framework.Based on the review findings, a five-stage closed-loop framework, zero energy building-physical AI robotics (ZEB-PAR), is derived: Perceive, Diagnose, Update, Optimize, and Verify.Technology readiness level (TRL) analysis revealed an asymmetry between the Perceive-Verify axis (mean TRL 7) and the Diagnose-Optimize axis (mean TRL 5), indicating that robotic data collection and re-inspection are nearly deployment-ready, whereas physics-informed AI and LLM-based optimization require further building-domain adaptation.