Xu Weng (翁旭)

xu009@e.ntu.edu.sg

I'm a PhD student at Nanyang Technological University in Singapore, where I am fortunate to be supervised by Prof. KV Ling and supported by Nanyang Research Scholarship. Priorly, I received a B.Eng. from Nanjing University of Aeronautics and Astronautics in 2015, as well as a M.Eng. from Beihang University in 2018, both in Electrical Engineering. I worked as an R&D engineer at Keysight Technologies from 2019 to 2020.

My research focuses on spatial awareness in the internet of things, especially user-oriented localization from a data-driven perspective.

CV  /  Scholar  /  Github

Research

A data-driven framework for correcting ranging errors, which is trained end-to-end using location-related loss. When combined with Euclidean distance field cost maps, NeRC can be trained using unlabeled data.

A satellite-wise MLP is designed to regress pseudorange errors from six satellite, receiver, context-related features derived from Android raw GNSS measurements. A novel method for labeling pseudorange errors are proposed.

An Augmented Reality (AR)-assisted framework to help exclude Non-Line-of-Sight (NLOS) satellites to improve GNSS localization using mobile devices.

A tool to log the relative location measurements from UWB and AR sensors mounted on iOS devices, as well as context-related data.

A framework unifying extended Kalman filter (EKF), factor graph optimization (FGO), and moving horizon estimation (MHE).

The first neural ranging correction framework trained end-to-end through a differentiable localization engine.

An open-source positioning toolbox for noisy Android raw GNSS measurements