Research made tangible.

The project develops, optimises and validates artificial neural networks designed to identify groups of asteroids that share a common origin.

Alongside the machine-learning work, I am exploring immersive visualisation methods for interrogating both real and synthetic training data, with the aim of making model behaviour and classification structure easier to inspect.

This creates a human-in-the-loop workflow in which statistical methods and visual reasoning support one another rather than operating as separate stages of the analysis.

VR2Dome builds on my doctoral work in digital-planetarium interaction and my professional experience in immersive data visualisation. It explores new methods for interacting with large-scale immersive data environments, with a focus on accessibility, collaboration, scientific utility and digital-twin approaches for immersive facilities.

The concept bridges the strengths of VR and the dome: a researcher can navigate and select data through an intuitive individual interface while a group shares the resulting view and analysis inside the theatre. This lowers barriers to scientific use without losing the dome's value for education and public engagement.

The current phase is exploring how VR2Dome could develop into a human-in-the-loop digital twin of a planetarium, linking a virtual representation of the dome to live or simulated system state and, potentially, bidirectional control for monitoring, rehearsal, operation and collaborative analysis.

Early proof-of-concept demonstrations were developed with IDIA and Iziko Planetarium in Cape Town and presented at data-visualisation workshops. A live demonstration at the 2019 Data to Dome Workshop was later highlighted in the President's Message of The Planetarian.

As lead developer, I work across the iDaVIE software suite. Its principal tools render scientific datasets in a room-scale immersive environment where users can navigate, select, inspect and edit structures directly in three dimensions.

The interaction model is designed to reduce the additional mental work involved in reasoning about 3D structures through conventional 2D interfaces, allowing researchers to focus more directly on scientific relationships and anomalies.

I also lead the iDaVIE-d prototype, which brings the rendering and interaction capabilities of the VR environment into the digital planetarium.

Although astronomy remains the primary application, the platform has also supported exploratory work with datasets from neuroscience, chemical engineering and astrobiology. iDaVIE-v is developed in collaboration with INAF-OAC.

SpaceDome combines real-time geospatial data, AI orchestration and spatial computing to support intuitive interaction with large Earth-observation datasets.

My role covers technical direction across the platform, including system architecture, data pipelines, visualisation workflows and the integration of individual VR exploration with collaborative digital-dome experiences.

The work also explores natural-language interfaces and AI-assisted navigation so that users can interrogate complex scientific information without losing the human role in directing and interpreting the analysis.

The study combined student responses from two university planetarium visits with detailed observational data. I designed the response instrument, created show content for the second visit and analysed both datasets using the Grounded Theory Method.

The student data produced a localised model centred on a spectrum of attentiveness that strongly influenced engagement. The observational data produced a second model describing the contextual factors that shape the planetarium teaching and learning space.

I then connected the localised models with Working Memory and Cognitive Load Theory to develop the Model for Curriculum Design in the Planetarium, or MCDiP: a systematic framework for shaping the digital planetarium as an effective teaching and learning environment.

As part of the Advanced Rayleigh Guided Ground Layer Adaptive Optics System, I led development of the Laser Alignment Telescope used to detect the initial positions of the guide lasers on sky and align them with the optical axis.

The work combined instrument design, construction, programming, installation and commissioning at the Large Binocular Telescope Observatory in Arizona.

The automated alignment approach was released as the open-source Collimating Smart Laser Alignment Algorithm, or ColSLAA, which uses LAT images to determine the pointing directions of the guide stars.