Interactive Data Fusion of Neural Radiance Fields for Industrial Facility Inspection in Virtual Reality

Ke Li, Susanne Schmidt, Tim Rolff, Reinhard Bacher, Wim Leemans, Frank Steinicke

Abstract: Real-world industrial facilities often present complex equipment that requires extensive inspection and maintenance for their operations. In this work, we present a virtual reality (VR) framework that supports virtual facility inspection and maintenance tasks by using neural radiance fields (NeRF) models to replicate, store, and visualize the appearance of complex industrial facilities. To overcome the performance bottleneck of VR NeRF rendering, we present two novel interactive data fusion techniques that can merge a NeRF model with its' corresponding CAD model through a mixed reality tunneling effect and contextual 3D NeRF drawing interaction. Technical benchmarking results and preliminary expert feedback are presented for the initial evaluation of our framework.

External IDs:dblp:conf/vr/Li0RBLS24