As a typical labor-intensive sector, the construction industry heavily relies on human workers. However, the current construction industry has to confront challenging issues of labor shortages and safety management. Using construction robots can effectively solve these problems and increase automation in project execution and management. Recent studies have reported that future intelligent construction methods will rely on long-term highly cooperative production models involving human-robot collaboration. This paper systematically reviews the foundational technologies and research frontiers of construction robotics, traces the developmental trajectory of next-generation intelligent construction robots, and identifies four typical human-robot collaboration modes through analysis of construction site characteristics: operator-controlled mode emphasizing human-led real-time interaction, commander-execution mode achieving automated execution of high-risk tasks through remote instructions, collaborative assistant mode combining machine efficiency with human flexibility, and human augmentation mode enhancing worker capabilities through exoskeleton robotics. The study further highlights current technical challenges including insufficient environmental adaptability, communication instability, and cross-platform collaboration difficulties, while suggesting priority research directions for future development in this field.