The integration of robotics and digital twin (DT) technologies is redefining automation in the construction sector, enabling safer, data-driven, and more efficient operations across the project lifecycle. However, current robotic simulation tools often lack interoperability with construction-specific workflows and provide limited high-fidelity, real-time visualisation of robotic behaviour in complex site environments. This paper presents a robot operation system (ROS) 2–Unity framework that unifies DT visualisation, robotic control, and simulation-based validation within a single modular architecture. The proposed system establishes bidirectional communication between ROS 2 nodes responsible for motion control, sensing, and environment updates and a Unity-based virtual environment that provides photorealistic rendering, physics-based interaction, and scene management. The framework supports three operational modes: DT visualisation for real-time synchronisation with physical robots, control training and algorithm evaluation within a simulated environment, and simulation and testing for safe pre-deployment validation. A construction and demolition waste sorting case study is used to demonstrate the framework’s applicability. A UR5e robotic arm with a Robotiq 2F-85 gripper and conveyor system is modelled in Unity, with waste objects dynamically spawned to emulate off-site sorting. Quantitative results, including communication latency and end-effector tracking performance, validate the framework’s suitability for closed-loop robotic manipulation in unstructured construction scenarios. The framework provides a lightweight, scalable, and extensible platform for advancing construction robotics research and DT development towards sustainable and circular construction practices.