ProfessorQuan Bai

Professor of ICT

Information and Communication Technology

  • Professor of ICT
    Information and Communication Technology
  • +61 3 6226 1536 (Work)
  • School of Information and Communication Technology, Information and Communication Technology, 456 Centenary Building, Sandy Bay Campus, TAS

RESEARCH SUMMARy

Agentic AI, multi-agent systems, machine learning, agent-based modelling, knowledge discovery.

FUNDED PROJECTS

  • GRANT
    AI-HYDRA: From Soil Moisture Data to Irrigation Decisions
    1 Aug 2026 - 31 Mar 2027
    People funded by this grant:
    • Edwards S,
    • Boucher C
    Project Total: $67,950; The additional investment is sought to translate the validated modelling framework from Project 2.2.002 into a practical irrigation decision-support capability for end users. Although the existing project has demonstrated strong scientific performance, it has not yet been converted into an operational workflow that growers, irrigation advisors, or sensor platform partners can use to interpret soil moisture data and support irrigation decisions. Through collaboration with Riverine Plains Inc. and testing on selected farms using existing multi-depth soil moisture datasets, this proposal will add value by packaging the existing modelling assets into a minimum viable decision-support prototype and validating its practical relevance for grower use. Expected benefits include improved interpretation of soil moisture data, more informed irrigation timing, reduced risk of over-irrigation, improved water-use efficiency, and better decision support for productivity and profitability at farm level. As the Soil CRC approaches completion, this proposal will help ensure that the existing investment in Project 2.2.002 is more fully realised through an operationally tested, end-user-oriented prototype rather than remaining only as research code and demonstration material. Funded by: CRC for High Performance Soils Ltd - CRC Funding ($67,950); University of Tasmania.
  • GRANT
    AlpineWatch Tasmania: A Crowd-Sourced Multimodal Citizen Science Platform for Real-Time Biodiversity Monitoring in Alpine and Subalpine Ecosystems Using AI-Powered Multi-Agent Analysis
    29 Jun 2026 - 29 Jun 2027
    People funded by this grant:
    • Sajjadi S,
    • Bai Q,
    • Yang W
    Project Total: $10,000; Tasmania's alpine and subalpine ecosystems, which include areas such as the Central Plateau, Mount Field, Cradle Mountain, and the Southwest wilderness, are home to a unique array of endemic species. These include cushion plant communities like Donatia novae-zelandiae and Oreobolus acicularis, as well as Pandani (Richea pandanifolia), the Snow Gum (Eucalyptus pauciflora), and various alpine invertebrates and vertebrates found nowhere else on Earth. These ecosystems are increasingly threatened by climate-driven warming, changes in snowpack dynamics, rising fire frequency, and the upward migration of subalpine species. However, baseline biodiversity monitoring in these areas is limited due to their remote and rugged terrain. To address this monitoring gap, the AlpineWatch Tasmania project has been created. This is a crowd-sourced, multimodal data collection platform that allows citizens, scientists, bushwalkers, and naturalists to submit geotagged photographs, audio recordings, and short video clips of the species, vegetation communities, and landscape features observed in Tasmania's alpine and subalpine regions. Each submission is accompanied by structured metadata, including details about the microhabitat type, substrate, weather, and elevation, and is instantly added to an accessible knowledge base for researchers. By significantly expanding the temporal and spatial resolution of biodiversity observations across Tasmania's alpine zone, this platform will: Enable near-continuous tracking of species distributions and population changes across previously unmonitored sitesProvide early detection of invasive species, such as Hieracium spp. and browsing ungulates, that threaten sensitive alpine habitatsGenerate longitudinal datasets that document climate-driven shifts in phenology and range contractions of endemic alpine taxaSupply conservation managers at the Tasmania Parks and Wildlife Service with community-generated evidence to guide adaptive management decisions Ultimately, this project enhances the resilience of Tasmanian alpine ecosystems by ensuring that the species and ecological processes within these areas are observed, documented, and understood at a scale beyond what individual researcher surveys can achieve. Funded by: Mountain Air Foundation - Research Grant Program ($10,000); University of Tasmania.
  • CONSULTANCY/CONTRACT RESEARCH
    UTAS and TasNetwork - AI Capability Uplift Project
    13 Mar 2026 - 1 Dec 2026
    People funded by this grant:
    • de Salas K,
    • Roehrer E,
    • Wong M,
    • Lewis I,
    • Merritt T
    Project Total: $50,000; The School of ICT at University of Tasmania will deliver a first-stage AI capability uplift program for TasNetworks, focused on building shared understanding, surfacing priority opportunities, and establishing a pipeline for future research, innovation, and engagement activities. This initial phase will also inform the development of a reusable catalogue of uplift activities that can support longer-term collaboration. The program will include: Executive Think Tank (co-facilitated): Co-design and facilitation of an executive-level think tank to help articulate priority problem and opportunity statements, explore strategic AI use cases, and align opportunities with organisational goals. Partnership Hackathon: Co-hosting a collaborative hackathon with TasNetworks and AWS to explore selected challenges through rapid prototyping, experimentation, and cross-disciplinary collaboration. PostHackerton Think Tank Curation Workshop (co-facilitated): Targeted facilitation to refine, prioritise, and shape identified problems into well-defined challenges suitable for progression into applied projects and hackathon activities. AI Demystification Session (in-person): An interactive session for staff focused on demystifying AI, building shared language and confidence, and clarifying realistic capabilities, limitations, and use cases. NavigateAI Platform Access: Provision of access to the NavigateAI platform to support early-stage AI literacy, shared understanding, and ongoing capability uplift across participating staff. Funded by: Tasmanian Networks Pty Ltd ($49,954); University of Tasmania.
  • GRANT
    Language Identification and Communication Project (LICP) - formerly LIPS
    National Emergency Management Agency1 Jul 2025 - 1 Dec 2026
    People funded by this grant:
    • Bai Q
    Project Total: $253,937; The LIPS Project has pioneered AI-powered software designed to bridge linguistic gaps, facilitating effective and reliable communication across various languages for specific purposes. Spearheaded by the Glamorgan Spring Bay Unit of the Tasmanian State Emergency Services (SES) and overseen by The Association for The Identification of Spoken Language Inc., this initiative aims to enhance communication capabilities. However, the current iteration of the LIPS system has limitationsit supports only a limited number of languages, and its accuracy falls short of meeting real-world application standards. To address these shortcomings, the project is focused on exploring novel machine learning algorithms to bolster the performance of LIPS, ultimately rendering it operationally robust and deployable. Funded by: National Emergency Management Agency - Grant-Disaster Ready Fund ($253,937); University of Tasmania.
  • GRANT
    Assessing the feasibility of integrating LLM and GenAI to enhance information accessibility across Natural Hazards Research Australia
    2 Jun 2025 - 30 May 2026
    People funded by this grant:
    • Bai Q,
    • Yang W
    Project Total: $65,909; In this project, we will undertake a Feasibility Study to evaluate the integration of a Large Language Model (LLM)-powered chatbot using Retrieval-Augmented Generation (RAG), with the objective of enhancing research project-related information accessibility across the Natural Hazards Research Australia. This feasibility study will explore three architectural options for implementing the RAG-based chatbot: External RAG Service: Leveraging a fully managed external service (e.g., Microsoft Copilot), requiring no Centre-specific development or infrastructure. Mixed RAG Service: Utilising an external LLM and embedding model via API, combined with a Centre-hosted vector database and Centre-developed RAG application layer. Internal RAG Service: Establishing a fully internal solution involving a Centre-managed, isolated DeepSeek LLM, Centre-hosted vector database, and a custom-built RAG service. The study will: Assess the technical feasibility, benefits, and limitations of each option. Identify potential opportunities and risks related to data security, maintenance, scalability, and user adoption. Provide a comparative analysis of projected costs and development/operational timelines over 3-year, 5-year, and 7-year horizons. The outcome will inform decision-making on the most appropriate and sustainable approach to implementing a RAG-based chatbot solution tailored to the Natural Hazards Research Australia's needs. Funded by: Natural Hazards Research Australia - Grant ($65,908); University of Tasmania.
  • GRANT
    Augmented Human Operations
    1 Jan 2024 - 31 Dec 2029
    People funded by this grant:
    • Chinthammit W,
    • Billinghurst M,
    • Dwyer T,
    • Cordeil M,
    • Bai Q
    Project Total: $1,578,647; The proposed research aims to explore the concept of augmented human operations where the human is working with machines. The study will focus on the four key areas of collaboration, training, digital tools, and literacy of the tools. The research will investigate how these areas can be integrated to create a more efficient and effective workforce. This research will examine the concept in various industries as we recognise an opportunity to enable Australian industry sectors to enhance their operations through augmented human operations to boost Australias economy. Our core objective is to empower human performance, capabilities, and safety with digital technologies to facilitate seamless communication and cross-interactions between human and digital entities. We aim to innovate by fusing the capabilities of AI and humans in the focused key areas via immersive VR and AR technologies. The proposed research will provide valuable insights into how augmented human operations can be leveraged to create a more efficient and effective workforce. The findings of this study will have significant implications for businesses and organizations looking to optimize their operations through the use of advanced immersive technologies. Funded by: CSIRO-Commonwealth Scientific & Industrial Research Organisation - Grant ($560,245); University of Tasmania.
  • GRANT
    Hiding in plain sight: Early identification of neurodegenerative disorders with computer analysis of face and eye movements
    1 Jan 2024 - 31 Dec 2028
    People funded by this grant:
    • Alty J,
    • Vickers J,
    • St George R,
    • Bindoff A,
    • Bai Q
    Project Total: $1,006,412; Alzheimers disease (AD) and Parkinsons disease (PD) start damaging the brain 15-20 years before symptoms begin. At the time of diagnosis, AD and PD are hard to treat because the damage is so severe. We need a way to detect AD and PD much earlier. We will develop a simple new computer test to detect early signs of AD and PD by analysing face and eye movements. Then people can start prevention earlier and scientists can research better treatments to improve peoples quality of life. Funded by: National Health & Medical Research CouncilNational Health & Medical Research Council - Grant-Ideas ($2,012,824); University of Tasmania.
  • CONSULTANCY/CONTRACT RESEARCH
    A data-driven machine learning solution and cloud service for close to real-time seaborne carbon accounting and reporting
    1 Jan 2023 - 31 Dec 2024
    People funded by this grant:
    • Fei J,
    • Kang B,
    • Bai Q,
    • Du B
    Project Total: $59,192; This project aims to develop an economical but highly accurate solution to monitor, measure, and report the seaborne carbon footprint of each container on board a ship at sea, by taking advantage of the data accumulated in the shipping company. The solution will be deployed as a cloud service by using e.g., Amazon Web Services (AWS) or other cloud service platforms, minimising the needs of investment on hardware and software. Data will be acquired through the shipping company, commercial providers (e.g., AIS), and public sources (e.g., meteorological data). With these datasets, state-of-the-art machine learning models will be developed to accurately quantify the fuel consumption and accompanying CO2 emissions of a ship at different speed, draft/displacement, trim, weather, and sea conditions. The seaborn carbon accounting and reporting can be achieved at both ship and container levels. A dashboard for carbon accounting and reporting will be developed for the shipping company. Funded by: La Trobe University; Solupia ($59,192); University of Tasmania.
  • CONSULTANCY/CONTRACT RESEARCH
    Virtual Database for Cloud-Based Data Services study
    1 Dec 2022 - 31 Dec 2023
    People funded by this grant:
    • Kang B,
    • Bai Q,
    • Amin MB,
    • Yang W
    Project Total: $120,000; The Australian Antarctic Data Centre (AADC) is managing an extensive amount of data collected from the Southern Ocean. The data are collected by different agencies and in various formats. To achieve better data management, discovery useful knowledge and provide accurate decision support, AADC seeks to increase the machine-readability of its data collections to facilitate data interoperability and integration, and to improve their usability for machine learning applications. This will require annotating data collections with machine-readable metadata describing data structures and meaning, and develop an effective automated metadata management mechanism, i.e., to develop a Virtual Database (VDB). This project will focus on building on technologies developed during the previous year to implement a functioning prototype system. The work will be divided into two principal goals: 1. Develop a data management module called VDB that virtually centralises the AADCs data by providing comprehensive and consistent access to data via a secure abstraction layer. This layer can further participate as an API to support extended usability and services such as Analytics and Visualisation. 2. Develop an AI-assisted meta-data management (MDM) component as an extension to VDB, which will provide a high-quality user experience for metadata entry. In addition, it will provide AI-driven assistance in identifying potential text entries to generate consistent metadata entries. Funded by: Australian Antarctic Division ($120,000); University of Tasmania.
  • GRANT
    Development of a non-invasive screening test to detect risk of Alzheimer's disease pathology
    1 Jan 2022 - 28 Feb 2027
    People funded by this grant:
    • Alty J,
    • Goldberg L,
    • Roccati E,
    • Lawler K,
    • Bhagwat A
    Project Total: $446,957; Alzheimer's disease (AD) starts damaging the brain >10 years before memory problems begin. To reduce dementia incidence, we must detect AD earlier so we can target prevention through early clinical trial recruitment and risk reduction (that could prevent 40% cases). With 10,000 available UTAS research participants and 200 patients from RHH and the community, we aim to develop a test to detect early signs of AD by analysing hand- AND speech- movements using a smartphone and machine learning algorithms. Funded by: Royal Hobart Hospital Research Foundation - Grant-Major Project ($446,957); University of Tasmania.
  • GRANT
    Sustainable Knowledge Model Towards Responsible AI
    1 Jan 2022 - 30 May 2025
    People funded by this grant:
    • Bai Q,
    • Amin MB,
    • Xu S,
    • Li Q
    Project Total: $5,000; Currently, governments/social organisations propose multiple principles and regulations to guide the ethics of AI models/products/services. Technology companies also have their policies to make alignment with these principles and regulations. Our project aims to enhance the realisation of these ethical values in the workflow of AI models. 1) Ethical KG - We process the regulation documents written by human language and transfer them to a machine-readable knowledge model. 2) Provenance of AI Workflow - On the other hand, all the factors of AI model, such as source data, algorithm, prediction outcomes will be recorded. 3) Compliance Check - The comparison between these records and ethical knowledge model will indicate whether the targeted AI model satisfies the corresponding ethical requirements. Expected Outcomes: By doing this, the ethical alignment will be realised directly in AI model level, rather than the policy level. It will increase the trusts on the products/services based on the AI model. Moreover, it is sustainable as regulation evolving. There is a trend that different regulations converge to some common values. The knowledge model generated from regulation documents can be self-updated. Eventually, it can be extended to a framework covering the whole processes of the AI products/services. Funded by: CSIRO-Commonwealth Scientific & Industrial Research Organisation - Scholarship-Data61 ($5,000); University of Tasmania.
  • GRANT
    Data fusion and machine learning for ship fuel efficiency analysis: a small but essential step towards green shipping through data analytics
    1 May 2021 - 31 May 2022
    People funded by this grant:
    • Du B,
    • Chen P,
    • Nikolova N,
    • Bhaskar P,
    • Fei J
    Project Total: $77,790; This project is to * Develop a data fusion methodology that pre-processes voyage report data, sensor data, weather archive data, and AIS data and combines these four data sets in a systematic way; * Develop some ship fuel efficiency models with machine learning approaches (piecewise linear/nonlinear regression, neural network, and random forest) that quantify the synergetic influence of sailing speed, displacement/draft, trim, sea conditions, and weather conditions on a ship's daily bunker fuel consumption; * Construct a computer (Python) code infrastructure for the above data fusion methodology and ship fuel efficiency models. Funded by: International Association of Maritime Universities - Grant ($77,790); University of Tasmania.
  • GRANT
    Giant Crab Enhanced Data Collection - Innovative approaches to enhance data collection in the Victorian, South Australian and Tasmanian Giant crab fisheries
    28 Jan 2021 - 30 Sep 2026
    People funded by this grant:
    • Jeavons T,
    • Hadley S,
    • Leon Leiva R,
    • Quigley B,
    • Yang W
    Project Total: $311,121; This project will address identified shortcomings in the quality of data currently available for Giant Crab stock assessment and fisheries management (Tas, Vic and SA). This project will look to develop a method for fishers to images of crabs on the vessel. The images will processed using Visual Intelligence methods to determine length, sex and unique ID for each crab. This data will be used in stock assessment models to make more accurate assessment. Hopefully this process will be rolled out across the Giant Crab fishery and a database of images can be collected and stored in a central repository. Funded by: Fisheries Research & Development Corporation - Grant; Department of Natural Resources and Environment Tasmania; South Australian Research and Development Institute; Victorian Fisheries Authority; University of Tasmania.
  • GRANT
    The future in our hands: Screening for preclinical Alzheimer's disease by analysing hand movements
    1 Jan 2021 - 31 Dec 2026
    People funded by this grant:
    • Alty J,
    • Vickers J,
    • Bai Q,
    • Bindoff A,
    • Roccati E
    Project Total: $899,782; Global dementia prevalence is rising. Alzheimer's disease (AD), the most common cause, has devastating effects on people's quality of life. AD has a preclinical (pre-AD) period of 10-20 years when brain pathology silently progresses before any cognitive symptoms appear. Current tests for pre-AD are invasive, costly and unsuitable for screening at population level. If we can't identify people with pre-AD, we can't offer them early interventions. This hinders our strategies to reduce AD prevalence. We will develop a new scalable test (HandTest) by combining two innovative ideas: hand-movement tests to detect pre-AD >10 years before cognitive symptoms begin; and computer vision so people can self-test? online at home. This unique approach builds on the recent discoveries that the posterior cingulate cortex (the first brain region to show abnormalities in pre-AD) is involved in hand-movement control, and that hand-movement patterns change in pre-AD. We will use exquisitely precise computer vision methods to automatically analyse movement data from thousands of participants, who complete the online test by simply opening and closing their hands repeatedly for 5 minutes in front of a standard computer, tablet or smartphone camera. The CI/AI team of experts in hand-movement analysis, pre-AD, computer vision, AD biomarkers and biostatistics has access to 3 well-phenotyped cohorts, >10,000 existing participants and a cutting-edge assay for a blood AD biomarker, ptau181. We will develop a HandTest algorithm to classify hand-movement data for pre-AD risk. In a longitudinal cohort we will determine HandTest's precision to prospectively predict 5-year risks of cognitive decline and AD. This is the critical first step for up-scaling HandTest nationwide via memory clinics of the Australian Dementia Network. With worldwide access to computers, tablets and smartphones increasing, HandTest is scalable globally, potentially transforming dementia prevention and management. Funded by: National Health & Medical Research CouncilNational Health & Medical Research Council - Grant-Ideas ($1,799,564); University of Tasmania.
  • CONSULTANCY/CONTRACT RESEARCH
    Virtual Database for Antarctic Data Management
    1 Jan 2021 - 15 Jul 2022
    People funded by this grant:
    • Bai Q,
    • Amin MB
    Project Total: $10,000; The Australian Antarctic Data Centre (AADC) is managing huge amount of data collected from the Southern Ocean. The data are collected by different agencies and in various format. To achieve better data management, discovery useful knowledge and provide accurate decision support, AADC seeks to dramatically increase the machine readability of its data collections to facilitate data interoperability and integration, and to improve their usability for machine learning applications. This will require annotating data collections with machine-readable metadata describing data structures and meaning, and develop an effective automated meta data management mechanism, i.e., to develop a Virtual Database (VDB). In this project, the student will study on related knowledge on knowledge representation, meta-data management, and VDB. The student will then be focus on the design a VDB, which can achieve automated meta-data management for AADC. Funded by: Australian Antarctic Division ($10,000); University of Tasmania.
  • GRANT
    Investigation into Dynamic Reallocation of Computing Resources
    10 Feb 2020 - 28 Aug 2020
    People funded by this grant:
    • Garg S,
    • Kang B,
    • Amin MB,
    • Bai Q
    Project Total: $75,000; The proposal for Whitepaper development project for Lockheed Martin Australia Attack Class Submarine Program EOI, FSP-CS-RT-80 Smart driven dynamic reallocation of computing resources based on compute demands Funded by: Lockheed Martin Australia Pty Ltd - Future Submarine Program ($75,000); University of Tasmania.