Real-time inspection in construction: Integrating ROS and APS for computer vision applications

A major challenge toward construction robotization is a lack of integration and communication between robots' data on construction processes and platforms that provide APIs and clouds. While previous studies have explored cloud-based integrations, this research uniquely combines Autodesk Platform Services (APS) with the Robot Operating System (ROS) to address this gap. This study expands to include APS and robot inspection, generating data through Computer Vision (CV), and it proposes a platform that combines ROS with APS. The proposed platform enables seamless interaction between users and robots in construction while addressing key challenges such as real-time data visualization and system scalability. It is practical for all kinds of robots. Still, we used a simulated Unmanned Aerial Vehicle (UAV) equipped with a camera in the Gazebo simulator to visualize automated and real-time inspections in APS Viewer. The trained model detects and counts bolts with a precision of 0.882 and a recall of 0.854, and the data is transmitted via Web Socket to APS for real-time visualization. To bridge telemetry data between the ROS node and APS, PX4 autopilot was used. The result demonstrates the feasibility of developing ROS on different web-based platforms, improving autonomous systems for saving cost, and time, and enhancing safety in construction projects.