This research develops reliable AI-powered drone systems to support New Zealand’s Predator Free 2050 initiative. By improving neural network calibration, uncertainty estimation, and robustness in challenging real-world conditions, the project aims to accurately detect invasive predators and better protect endangered native bird species.

This PhD uses brain-inspired AI to decode vision from neural data. Using human fMRI (24 hours of Doctor Who) and monkey electrophysiology, signals are transformed into 2D brain maps to improve reconstruction. The model learns receptive-field structure, compares contributions of V1/V4/IT, and aims for efficient, interpretable decoding with applications to neuroscience and BCIs.

Traditional neural networks are powerful but difficult to interpret and vulnerable to small input changes. This research develops wavelet-based neural networks with provable stability guarantees, extending the scattering transform to texture modeling. The approach reduces feature complexity while improving interpretability, enabling more reliable and mathematically explainable AI systems.

Marine debris severely harms ocean ecosystems, yet most cleanup focuses only on floating waste. This research develops underwater robots equipped with specialized computer vision to detect and remove submerged trash. By training algorithms for challenging underwater conditions, the work enables safer, scalable cleanup operations and supports long-term ocean preservation.

Industrial combustion residue can strengthen concrete but varies in impurity content. This research uses X-ray imaging and computer vision to identify and quantify impurities in residue particles. The results help cement manufacturers optimize material use, improving quality, reducing costs, and supporting sustainable recycling of industrial waste.

My research uses field images to predict crop yield, leveraging machine learning techniques to extract patterns and features correlating yield.  These features include plant health indicators, growth stages,  or canopy coverage. I am particularly interested in using these features to develop models  that improve the accuracy of yield prediction, helping farmers make  data-driven decisions. My approach considers temporal changes in the crop, capturing how its characteristics evolve. My work contributes to precision agriculture, a field that seeks to optimize resource use, increase productivity, and promote sustainability in farming. My research has the potential to transform traditional agricultural practices by integrating advanced AI methods.

This talk explains research that teaches legged robots how to walk reliably using machine learning, computer vision, advanced control theory, and Lyapunov-based safety guarantees. By improving robot stability on complex terrain, the work moves us closer to versatile, household multi-purpose robots capable of performing everyday chores safely and independently.

Mel-AI is an artificial intelligence system designed to assist pathologists in distinguishing melanoma from benign moles. By training computer-vision models on 520 cases, the system reached 96% accuracy and interpretable outputs. It offers scalable, objective quality assurance, reducing misdiagnosis risk and improving melanoma detection in high-incidence countries like Australia.