Stephen Hausler

Queensland University of Technology, Commonwealth Scientific and Industrial Research Organisation

Papers

Total Citations

H-Index

About

Stephen Hausler is a leading researcher in visual place recognition and 3D scene understanding for robotics and autonomous systems. His work centers on developing robust algorithms that enable machines to recognize locations despite dramatic changes in appearance, viewpoint, and environmental conditions. Hausler’s most impactful contribution is **Patch-NetVLAD** (2021, 14+ citations), which introduced a novel multi-scale fusion of locally-global descriptors for place recognition. This work cleverly combines the robustness of global descriptors with the precision of local features, setting a new benchmark for performance under challenging real-world conditions. He has also advanced bio-inspired multi-scale fusion techniques (2020) and, more recently, tackled the registration of neural fields with **Reg-NF** (2024), enabling efficient alignment of implicit 3D scene representations. By bridging classical geometric methods with modern neural field approaches, Hausler’s research directly impacts the reliability of long-term robot navigation and autonomous mapping. His work is essential reading for anyone building systems that must operate reliably in a changing world.