Animation Authoring for Industrial VR Applications: Integrating Deep Learning and Motion Paths

Enhancing industrial workflows with virtual reality (VR) is a key part of pushing modern manufacturing into the next stage. A major challenge in this transformation is the dependence on third-party services for critical tasks such as 3D animation, preventing companies from taking complete ownership of the creative and operational processes. This PhD thesis addresses this issue by empowering industrial experts in VR to create object animations and exploded view animations themselves, contributing to a more autonomous and integrated approach to VR in industrial settings.

In this thesis, a VR animation tool is designed and developed to simplify the process of 3D object animation through the principles of performance animation and motion paths. Furthermore, the automatic generation of exploded view animations in VR is explored, with real-time optimization of previous approaches as well as a first ever approach based on machine learning. By making two point cloud datasets and deep learning methods publicly available, and incorporating them into a VR animation tool, this research contributes to the advancement of both VR and 3D animation. It demonstrates the potential of these technologies to transform industrial workflows, making complex animation tasks more accessible and promoting innovation in manufacturing practices.