Hello everyone, you can find my introduction here. I am employed by Mid Sweden University (MIUN), and title of my research is “Augmented Telepresence for Interactive Remote Actions”. In this project I aim to develop methods which can present a remote scene in real time such as it can enable operating a remote machine possible.
ESR11 Manu Gond
An augmented telepresence pipeline can be split in 3 parts:
1 – Capturing devices
2 – Pre-processing & transmission
3 – Post-processing & presentation.
Currently I am focusing on part 1. Generally having a detailed capture of scene requires multiple cameras and sensors, but as a downside, it increases both computation complexity as well as bandwidth requirements. Thus, minimizing number of capture devices can improve the bandwidth requirement and complexity of post processing thus improving step 2 and 3. For this I am looking into Deep Learning based methods to reconstruct surrounding views from a single or sparse number of input views. I am more focused on Light Field reconstruction to achieve this. For testing the hypothesis I am utilizing some synthetic dataset rendered using 3D design software.
Academic Secondment
I have a six month secondment at Technische Universität Berlin (TUB), Germany from January 2023. I will be focusing on optimizing developed telepresence systems for real time usage and formulating some user studies. The goal is to finally have this system on actual testbed where user study on Quality of Experience can be performed.