Joshua Knights
I am a Post-Doctoral Research Associate in Robotic Perception at the Australian Centre for Robotics, The University of Sydney. My research focuses on 3D scene understanding for robots operating in unstructured, natural environments — including LiDAR and visual place recognition, cross-modal re-localisation, and adapting learned perception models to novel domains.
I completed my PhD at Queensland University of Technology in a joint program with CSIRO’s Data61, supervised by Peyman Moghadam, Clinton Fookes and Sridha Sridharan, and received the 2025 QUT Outstanding Doctoral Thesis Award for my thesis, Bridging Domain Gaps in 3D Scene Understanding. In 2024 I was a Research Engineer at the Smart Robotics Lab, Technical University of Munich, developing perception pipelines for autonomous drones as part of the EU Horizon project DigiForest.
News
- 2026Started as a Post-Doctoral Research Associate in Robotic Perception at the Australian Centre for Robotics, The University of Sydney.
- 2026Awarded the 2025 QUT Outstanding Doctoral Thesis Award for my PhD thesis, “Bridging Domain Gaps in 3D Scene Understanding.”
- 2026Our paper WildCross, a benchmark for place recognition and metric depth estimation in natural environments, was accepted to ICRA 2026.
- 2025Completed my PhD at QUT, a joint program with CSIRO’s Data61, supervised by Peyman Moghadam, Clinton Fookes and Sridha Sridharan.
- 2025Our paper SOLVR, a submap-oriented LiDAR-visual re-localisation pipeline developed at the Smart Robotics Lab, TU Munich, was accepted to ICRA 2025.
- 2024Worked as a Research Engineer at the Smart Robotics Lab, Technical University of Munich, developing computer vision pipelines for autonomous drones as part of the EU Horizon project DigiForest.
- 2024Our paper WildScenes, a benchmark for 2D/3D semantic segmentation in natural environments, was published in the International Journal of Robotics Research.
Selected Publications
All publications →
WildCross: A Cross-Modal Large Scale Benchmark for Place Recognition and Metric Depth Estimation in Natural Environments
IEEE International Conference on Robotics and Automation (ICRA), 2026
SOLVR: Submap Oriented LiDAR-Visual Re-Localisation
IEEE International Conference on Robotics and Automation (ICRA), 2025
GeoAdapt: Self-Supervised Test-Time Adaptation in LiDAR Place Recognition Using Geometric Priors
IEEE Robotics and Automation Letters (RA-L), 2024
Wild-Places: A Large-Scale Dataset for Lidar Place Recognition in Unstructured Natural Environments
IEEE International Conference on Robotics and Automation (ICRA), 2023