ROBOTIC SOLUTIONS FOR 3D SPACE ANALYSIS
AIM
This workshop ran in collaboration with Noumena and IAAC where we were introduced to a environmental scanning and data processing workflows which combined the use of photogrammetry and point-cloud data. The site of the study was at IAAC’s new building which is yet to be constructed. From this data we would develop a deep learning model using robo-flow that would be able to analyze the current health of the roof and truss supports. Given the limited time of the workshop we treated this as an exercise in how to meaningfully extract and communicate findings through data in the representation of a point-cloud. this lead us to develop a web landing of the point-cloud that the user can interact and move within – this model would contain statistics linked to coloured points within the digital model.
Data workflow
Drone scanning path
Point Cloud (Exterior)
Point Cloud (Interior)
Image matching process
Deep learning data set methodology
Labeling in Robo-flow
Results from the training matrix
Trained model tagging processed images
Point cloud data on web landing page
