DEEP-BIM: Dynamic Enhanced Environment Perception - Building Information Models

Toyon Research Corp. proposes development of a software prototype that processes 3D point cloud data and other data such as images to generate building information model (BIM) updates in near-real-time. The software includes algorithms for 3D computer vision processing and emerging AI/ML models for 3D point cloud processing using deep neural networks. The software will detect, segment, classify, and model relevant building components and objects within buildings and underground facilities. The software will be developed as Robot Operating System (ROS) modules enabling convenient integration with other systems, including integration in the GovernmentÆs E2MAPS system. Initial integration will be performed early in the project, and capabilities will be added throughout to form a minimum viable product, to add additional capabilities such as extended infrastructure modeling, and to incorporate feedback from end user representatives.