Research
Brief summaries of some of the Lab’s recent research can be found below. Feel free to contact the team if you’d like any more information on any of our research.
MapTalk
Previously called “Chat with your data”, the MapTalk project combines tried-and-tested geospatial queries and spatial analytics with new generative AI technologies, using a technique known as orchestration, or more recently RAG (retrieval augmented generation).
The approach shows huge promise for making automated access to authoritative spatial data and trustworthy spatial tools much simpler and easier for all. Leveraging the text processing strengths of generative AI together with spatial data and analytics holds the potential for anyone, not only spatial data analysts and GIS technicians, to easily and rapidly access much more of our authoritative spatial data for everyday decisions.
Demonstrating the power of this approach, the research team built an AI chat interface to CLUE, the authoritative Census of Land Use and Employment, made available publicly by the City of Melbourne. More recently, the work has developed in to a funded AEA Ignite project in collaboration with industry partners Veris and Surround Australia.
Dynamic Vicmap
The increasing variety and volume of spatial data streams present major challenges to the maintenance of foundation spatial data.
The Dynamic Vicmap project, a collaboration between RMIT University researchers, the SmartSat CRC, Victorian Government, and FrontierSI, proposed a fundamental shift towards a semantic, knowledge-based approach to the management of foundation spatial data updates.
The team developed a proof-of-concept demonstrator that can automate the process of semantic enrichment, integration, interoperability, and dynamic updates of spatial data. The demonstrator was tested with statewide authoritative data and new data streams, including Vicmap hydro, Vicmap property, and machine-learned flooding data.
The results demonstrate the potential for the approach to assist in easier integration of multiple data sets and layers, associated provenance and metadata, and smarter, simpler queries for dynamic foundational spatial data, in Victoria and internationally.
The Victorian Government has developed a suite of urban Priority Precincts such as Fishermans Bend. The FILTER project is pioneering a new framework for evaluating outcomes in precincts.
The approach innovates by combining robust, place-based evaluation with quantitative metrics, data-driven benchmarks, and digital tools. FILTER is designed to apply to a range of current and future precincts, with a demonstrator to be developed for Australia’s largest urban renewal project, Fishermans Bend in Victoria.
The FILTER project is a collaboration between RMIT University, including Geographic Knowledge Lab researchers, Victorian Government Department of Transport and Planning (DTP), and the iMove CRC.
The project outcomes will have implications for understanding progress in precincts statewide, nationally, and internationally.
NEXUS
NEXUS addresses the problem of how to make sense of the deluge of physics-based and human sensor data for more timely and accurate intelligence.
The project developed a real-time, streaming geospatial intelligence analytics system capable of support a very wide range of queries, with a particular focus on movement analytics. The system is compatible with standard reference system architectures in defence and industry. Building on this foundations offers the potential to support a widening range of more sophisticated queries, including causal reasoning about the intentions of moving objects.
More information on the NEXUS project appears in publications including:
Duckham, M., Gabela, J., Kealy, A., et al. (2022). Explainable spatiotemporal reasoning for geospatial intelligence applications. Transactions in GIS, 26, 2455–2479. https://doi.org/10.1111/tgis.12939
Duckham, M., Gabela, J., Kealy, A., et al. (2023). Qualitative spatial reasoning with uncertain evidence using Markov logic networks. International Journal of Geographical Information Science, 37(9), 2067–2100. https://doi.org/10.1080/13658816.2023.2231044
Update and revision
We rely every day on accurate and up-to-date road network data. But maintaining accurate global datasets in a dynamic world is a major challenge.
Machine learning can help. But machine learning struggles when training data is sparse or limited. To tackle this problem, we designed a new algorithm based on analysis of traveled routes, instead of machine learning.
Our algorithm searches for unexpected behaviors, or movements that are inconsistent with network data. Working with Here Technologies, we are investigating ways to apply these algorithms to more automated and reliable update and revision of road network data at a global scale.