We're offering a fully funded PhD position in AI-driven vestibular testing. Position closes 22 April 2025, Sydney Time. For full details see here.

We're recruiting a 2-year Postdoctoral Research Associate in Introspective Robotic Perception. Position closes 17 April 2025, Sydney Time. Details at the University Job Site here.

We received an NVIDIA Academic Grant for our project "Relightable Radiance Fields for Robotic Inspection in Dynamic Illumination". The GPUs they're sending will help us do even more with NVIDIA's cutting-edge ray-tracing and deep learning capabilities. Big thanks to the team for putting this together, especially Jason Lai, James Gray, Oliver Yan, and Jack Naylor!

We have two papers at WACV! In Oral Session 9.3 Oliver Yan is presenting TaCOS: Task-Specific Camera Optimization with Simulation. We address discrete, categorical, and continuous-domain design variables as well as episodic, non-differentiable cost functions. Our approach is faster than prior work and beats state of the art off-the-shelf cameras.

At the WACI Workshop Nikolai Goncharov is presenting LFSAM: Segment Anything in Light Fields for Real-Time Applications via Constrained Prompting. We bring the SAM 2 foundation model into the light field domain without retraining. Our approach is faster than naive subaperture segmentation and leverages the light field structure for view consistency.

Check the project page links above for papers and code!

It's been a big final month of the year for the Robotic Imaging Lab!

We received an NVIDIA Academic Grant for our project "Physics-Based Simulation for Robotic Vision in Extreme Environments". The GPUs they're sending will help us leverage NVIDIA's cutting-edge ray-tracing and deep learning capabilities.

Congrats to PhD candidate Bina Rajan who received first prize for her submission to the SafetyCulture Spot Competition along with Michael Somerfield. Take a moment to watch their fantastic video! The Spot Mini they've been awarded will come in handy in our ARIAM Hub infrastructure inspection and management research.

Our student Jack Naylor has been awarded a Fullbright Future Scholarship! Jack will be visiting my colleague and friend Matthew O'Toole and his group at the Carnegie Mellon University Robotics Institute developing exciting new ideas using radiance fields for multi-modal sensing.

We have received an ARC Discovery Grant on "Introspection for Resilient Robotic Perception in Challenging Environments", with Stefan Williams at USyd and Thierry Peynot at QCR. This 3-year program will advance how robots reason about and adapt to their own perceptual limitations, enhancing resilience in challenging environments such as murky water and extreme weather where state-of-the-art approaches fail.

Finally, the group have received a bit of an early Christmas present in the form a new Agilex Scout Mini robot. It's a pretty capable and nimble platform, and we look forward to making it see and interact with the world in new ways, from photons to actions!

Check it out, our work appears in Blender 4.3! The new features are covered in this video.

Alexandre Cardaillac lead this work introducing new physics-based volume scattering phase functions into Blender 4.3. This was driven by our need to accurately simulate imagery from underwater robots. As you can see in the video, it has utility in all sorts of scattering environments including fog, clouds, and even interstellar dust! We hope it's useful in a wide range of scientific as well as artistic applications!

This work was conducted as part of the ARIAM Research Hub in collaboration with industry partner Advanced Navigation.

Congrats Jack Naylor for being one of five speakers at the IAC2024 NextGen Plenary in a few days!

We were featured in a set of ABC news articles exploring aspects of robot vacuum security: Radio piece "Robot vacuums spying on Australians", article "We hacked a robot vacuum...", and article "...robot vacuums collect photos and audio to train AI".

Welcome to our new team member, James Gray. James is a postdoctoral researcher working with ARIAM Hub member Trendspek on robotic vision for high-fidelity drone-based 3D reconstruction. Watch this space for more as his work develops!

A big welcome to our two latest team members, Jason Lai and Connor Langford, starting their PhDs in change detection and exoplanet detection. Watch this space for more as their work develops!

Closing 10th Jun Sydney time: 3-year Postdoctoral Research Associate in 3D reconstruction of reflective objects. Details at the University Job Site here.

ICRA 2024 in Yokohama was a blast, great to see such strong attendance from Australian robotics colleagues. If you missed us, here are the links to BuFF Burst Feature Finder (RA-L/ICRA), Adaptive Keyframe Selection for Online Iterative NeRF Construction (RoboNerF workshop), and Towards Event-Based Satellite Docking: A Photometrically Accurate Low-Earth Orbit Hardware Simulation (HERMES workshop).

We're excited to introduce TaCOS: Task-Specific Camera Optimisation with Simulation. We combine procedural environments with clever optimisers to design better cameras, supporting continuous, discrete, categorical & interdependent variables. Preprint/code here.

Our privacy-preserving camera was featured on Channel9 news featuring myself and Adam Taras. This was a collaboration with Niko Suenderhauf and Peter Corke at QUT Centre for Robotics. Project page here.

Check out the latest collaboration between Adam Taras and myself from USyd and Niko Suenderhauf and Peter Corke from QUT: "Inherently privacy-preserving vision for trustworthy autonomous systems: Needs and solutions", accepted to the Journal of Responsible Technology. We argue for a class of cameras that are more like optical-analogue computers: they protect your privacy by never measuring images. Preprint, dataset, and a reconstruction challenge are at the project page here.

Still buzzing from RVSS 2024 last week. Congrats to Robotic Imaging Lab member Bina Rajan and team for winning the workshop challenge!

Welcome to our latest team member, Jesse Mehami who has started his Postdoctoral role in the SmartSat CRC ISAM project. Jesse joins us from the Robotics Institute at UTS where he worked with noval and multi-camera systems.

A great way to close off 2023 was by attending SIGGRAPH Asia. Always a stimulating time!