This research aims to initiate an automated process deploying high-precision 3D sensing and online (embedded) visual learning engines that operate in natural environments and are able to adapt in a holistic lifetime manner to changes during operations. It targets establishing a groundbreaking technological concept for 3D machine vision as an interdisciplinary initiative linking cost effective and reliable depth sensing with advanced computer vision and embedded artificial intelligence to tackle challenges posed by time-varying visual scene perception in natural environments. The combination of both recent breakthrough developments in 3D sensing (i.e. solid state LiDAR) and in mobile computing (i.e. embedded neural computing engine) is the basis for this research to tackle real-time monitoring and visual analysis using a network of tiny, low-cost and low-power smart 3D cameras, towards digital management services for operations and processes. This research supports building an internationally relevant R&D team on 3D machine vision towards enabling digitalization capabilities. In the first period, we have built the project teams at the three institutions for collaboration. At UiS, we have recruited a PhD student who started late April 2022 and we have also assigned a co-supervision at UiS. Both NORCE and Tohoku University have setup a project team consisting of a PI and scientists. Furthermore, we were working during this period on survey paper tackling techniques on online visual learning. Moreover, we are elaborating on different solid state lidars that can be integrated into the 3D vision system. Unfortunately, we were not able to acquire lidars yet due to the world crisis hitting the supply chain. However, we have identified an adequate provider who should be able to deliver it in Q1 2023. Finally, we have made a keynote talk at the Vision Expo in Stuttgart (the world largest Exhibition on Machine Vision) in October 2022.
The overall objective of this research is to give up the currently manual and time-consuming operational processes with well-composed digitalization using real-time 3D vision and learning processes. This research aims to initiate an automated process deploying high-precision 3D sensing and online (embedded) visual learning engines that operate in natural environments and are able to adapt in a holistic lifetime manner to changes during operations.
It targets establishing a groundbreaking technological concept for 3D machine vision as an interdisciplinary initiative linking cost effective and reliable depth sensing with advanced computer vision and embedded artificial intelligence to tackle challenges posed by time-varying visual scene perception in natural environments.
The combination of both recent breakthrough developments in 3D sensing (i.e. solid state LiDAR) and in mobile computing (i.e. embedded neural computing engine) is the basis for this research to tackle real-time monitoring and visual analysis using a network of tiny, low-cost and low-power smart 3D cameras, towards digital management services for operations and processes. This research has a potential to a radical change in many parts of the value chain in the industry and society and selected scenarios will be designed for impact evaluation and opening up integration opportunities. This activity fits the key strategical development of NORCE and its perspective to be number 1 in Norway within the next 5 years in machine vision innovation.
This research supports building an internationally-relevant R&D team on 3D machine vision towards enabling digitalization capabilites. NORCE has the role to develop Norway as leading hub with excellence and high-level of expertise on innovation for 3D machine vision as an enabler for digital processes and services.
