Dr Alwyn Mathew

Research Interests

My research focuses on 3D computer vision for infrastructure. I adapt and improve mobile scanning systems, including their software, to capture accurate 3D geometry of buildings, roads and hydropower assets. I then build algorithms that turn this data into geometric digital twins. My work spans point cloud segmentation, relationship graph construction, automated progress monitoring, and geometric deep learning for large-scale built assets. Earlier work covered monocular and self-supervised depth estimation. My broader goal is to make high-quality digital twins faster, cheaper and more repeatable to produce across the infrastructure lifecycle.

Research Project

This project creates a Local Roads dataset to enable the integration of Local Roads into the Digital Roads initiative, allowing future developments to be applied to both the SRN and Local Roads concurrently. I am leading the adoption of the current 3D scanner to support this purpose. Cambridge city centre will be scanned with the in-house scanner and turned into a 3D textured mesh. 

Biography

Alwyn Mathew is a Research Associate in the Department of Engineering at the University of Cambridge, working with Professor Ioannis Brilakis. His research centres on 3D computer vision, with the aim of improving mobile 3D scanning and developing algorithms that generate digital twins of infrastructure. 

Alwyn completed his PhD in 3D computer vision at the Indian Institute of Technology (IIT) Patna, advised by Dr Jimson Mathew. His doctoral work addressed monocular depth estimation, including self-attention architectures for unrectified video sequences and novel loss functions for depth prediction. He also studied the vulnerability of depth estimators to adversarial attacks. He then moved to the Division of Imaging Science and Technology at the University of Dundee as a Postdoctoral Research Assistant. There he worked with Dr Luigi Manfredi on vision systems for soft endorobots, developing self-supervised depth estimation for the navigation system. This work received the Amazing Technology Session award at EAES 2023. 

Alwyn joined Cambridge in February 2023 and has since worked on several large digital twin projects. On BIM2TWIN, a digital building twin platform for construction management, he adapted the group’s in-house mobile scanner into a handheld configuration to capture the 3D geometry of buildings under construction, and developed automatic progress monitoring from the resulting point clouds. On OMICRON, a digital twin platform for road assets, he adapted the scanner for vehicle-mounted capture of road infrastructure and improved its software. On D-HYDROFLEX, which develops digital solutions to improve the sustainability and flexibility of hydropower assets, he develops algorithms that automatically generate geometric digital twins of dams from point clouds, using deep learning for segmentation and relationship graph construction. 

Alwyn is also a member of the Digital Roads (DRF) team. He leads the Local Roads dataset project (DR LA), which aims to bring local roads into the Digital Roads initiative so that future developments can be applied to both the Strategic Road Network (SRN) and local roads at the same time. He is leading the adoption of the project’s in-house 3D scanner for this work. Cambridge city centre will be scanned with the in-house scanner and turned into a 3D textured mesh.