Telstra-UOW Hub for AIOT Solutions

In response to Australia's rising challenge of recyclable contamination, this project has developed an innovative solution leveraging advanced computer vision and deep learning techniques. Partnering with the Wollongong City Council and REMONDIS, the project aimed to automate the detection of contaminants in recyclables, freeing truck drivers from manual monitoring tasks. By analysing real-time video footage from bin-emptying processes, implemented deep neural network algorithm identified various plastic bag contaminants, providing precise location-based data. This transformative approach not only streamlines waste management operations but also empowers targeted interventions to enhance recycling efficacy and reduce landfill contributions.

Related Publication

1. Iqbal, U., Barthelemy, J., Perez, P. and Davies, T., 2022. Edge-computing video analytics solution for automated plastic-bag contamination detection: a case from remondis. Sensors, 22(20), p.7821.

Floodborne debris poses significant challenges during flooding events, often leading to infrastructure damage and altered flow dynamics. Traditional flood models struggle to account for debris impacts, largely due to the complex nature of debris movement and lack of automated identification solutions. Partnering with AVCON, our initiative harnesses cutting-edge computer vision and deep learning techniques to develop a system capable of detecting and classifying various debris types from video footage. Leveraging state-of-the-art neural network architectures and transfer learning methodologies, our solution aims to provide real-time insights, aiding flood management agencies in proactive mitigation strategies.

Related Publication

1. Iqbal, U., Riaz, M.Z.B., Barthelemy, J., Hutchison, N. and Perez, P., 2022. Floodborne Objects Type Recognition Using Computer Vision to Mitigate Blockage Originated Floods. Water, 14(17), p.2605.

Successfully addressing the challenges of blockages at cross-drainage hydraulic structures (e.g., culverts), StopBlock implemented and developed innovative AI-oriented solutions. By harnessing state-of-the-art AIoT technology, including cutting-edge computer vision and edge computing powered by NVIDIA Jetson, our solution provides real-time assessment of culvert blockage statuses. Through the integration of advanced deep learning classification, detection and segmentation models, the project achieved precise detection and classification of blockage, ensuring timely alerts and optimised flood management strategies. Committed to privacy and transparency, the system transmitted only metadata, upholding data compliance while bolstering infrastructure resilience. Furthermore, the project has pioneered the creation of a comprehensive culvert dataset, contributing to collaborative advancements in stormwater management practices.

Related Publications

1. NVIDIA Technical Blog -- https://developer.nvidia.com/blog/an-aiot-solution-for-visual-blockage-detection-at-culverts/
2. Iqbal, U., Barthelemy, J., Li, W. and Perez, P., 2021. Automating visual blockage classification of culverts with deep learning. Applied Sciences, 11(16), p.7561.
3. Barthelemy, J., Amirghasemi, M., Arshad, B., Fay, C., Forehead, H., Hutchison, N., Iqbal, U., Li, Y., Qian, Y. and Perez, P., 2020. Problem-driven and technology-enabled solutions for safer communities: The case of stormwater management in the illawarra-shoalhaven region (nsw, australia). Handbook of smart cities, pp.1-28.
4. Iqbal, U., Barthelemy, J., Perez, P., Cooper, J. and Li, W., 2023. A scaled physical model study of culvert blockage exploring complex relationships between influential factors. Australasian Journal of Water Resources, 27(1), pp.191-204.
5. Iqbal, U., Barthelemy, J. and Perez, P., 2022. Prediction of hydraulic blockage at culverts from a single image using deep learning. Neural Computing and Applications, 34(23), pp.21101-21117.
6. Iqbal, U., Bin Riaz, M.Z., Barthelemy, J. and Perez, P., 2022. Prediction of hydraulic blockage at culverts using lab scale simulated hydraulic data. Urban Water Journal, 19(7), pp.686-699.
7. Iqbal, U., Bin Riaz, M.Z., Barthelemy, J. and Perez, P., 2023. Quantification of visual blockage at culverts using deep learning based computer vision models. Urban Water Journal, 20(1), pp.26-38.
8. Iqbal, U., Bin Riaz, M.Z., Barthelemy, J. and Perez, P., 2023. Artificial Intelligence of Things (AIoT)-oriented framework for blockage assessment at cross-drainage hydraulic structures. Australasian Journal of Water Resources, pp.1-11.

Clearlight Saunas, a manufacturer of infrared saunas, have approached the Hub seeking capabilities to create infrastructure surrounding capturing data related to physiological parameters of sauna users and implementing Artificial Intelligence to determine possible health-benefits of sauna use. This project was completed 2023. The project delivered:
• An evaluation of commercially available sensors suitable for measuring environmental and physiological data during sauna use;
• Development of custom prototype sensors where required to supplement commercially available sensors;
• Creation of a data system to take raw sensor data into a consumer-facing dashboard or alert system; and
• Integration of the above mentioned system into a prototype sauna.

The first phase of the HE Silo project is almost complete. The project addresses a pivotal challenge globally, and in regional Australia's grain industry, which contributes over $13 billion annually but faces up to 5% grain spoilage due to post-harvest degradation in silos.

Globally the Post harvest degradation is reported as high as 30% in the emerging countries, where protecting the harvested grains is extremely important for feeding the world, even a 1% improvement to Post Harvest losses, would and many millions of tonnes of available grains for consumption back into the system.

The primary cause is moisture and temperature fluctuations from inadequate venting, leading to spoilage issues like mould, insect infestations, and even combustion. To counter this, a prototype autonomous sensing and actuation system is being developed to mitigate moisture buildup in silos, aiming to revolutionise grain storage and reduce spoilage.

The smart venting solution autonomously detects and addresses moisture buildup, enhancing production efficiency and potentially unlocking significant economic benefits. Significant progress has been made, with a developed electronic and control system capable of intelligently managing a system to combat moisture buildup, alongside implementing LoRaWAN technology for real-time monitoring.

Efforts are underway to extend sensor lifespan through SMART power management. A second phase of this project is in development, with plans to implement the system in a full-size grain silo to further test and advance the solution.

The Technology and Women’s Safety project is part of the iMove CRC. This research is to assess the opportunity to detect behaviours that threaten or imply a threat to women’s safety and develop tools that do so.

The project is being led by the Queensland University of Technology (QUT) with UOW (through SMART) a collaborating party. The final report was delivered in December 2023.

This project aimed to alleviate commuter frustrations due to peak-hour road congestion, offering a more streamlined and effective public transport system.

The project culminated in a detailed report with five key recommendations:

• Development of an Integrated Multi-Modal Services Plan for the Illawarra that includes seamless interchange between modes
• Customer-centred design thinking should be utilised to develop the Services Plan, which prioritises an intra-regional Illawarra commuter focus
• Implementation of the Illawarra Integrated Multi-Modal Services Plan, including changes to the current operator service plan
• Provision of Service-driven infrastructure improvements to support the Illawarra Integrated Multi-Modal Services Plan implementation, such as:
Additional East West Link for Shellharbour City Centre Hub
• Additional M1 alignment in Services Plan for Wollongong CBD Commute Hub Integrated ticketing system across all transport modes, utilising Opal-enabled services.

The full report, and presentation that was delivered to key stakeholders can be found here:
https://www.rdaillawarra.com.au/projects/infrastructure/30-minute-city/

There was some media generated on this project with local radio and newspapers picking it up https://www.facebook.com/illawarrarda/

https://www.illawarramercury.com.au/story/8208535/the-30-minute-city-are-we-just-dreaming/

https://www.illawarramercury.com.au/story/8307625/public-transport-is-solution-to-gridlocked-roads-says-business-group/

The iOyster project is a collaboration between SMART, Oceanfarmr and NSW Department of Primary Industries.

We are using sensor technology in oyster grow-baskets to de-risk oyster farming. The device will provide early detection of harmful heat and biofouling to help prevent catastrophic losses in periods of extreme heat and drought.

Funded by the Federal Future Drought Fund.

There is growing interest in biodiversity credits, but aquaculture does not yet have a way to collect evidence for investors.

This project is developing a methodology and prototype system of low-cost underwater cameras, water quality instruments and AI for gathering evidence of biodiversity on oyster farms. This will help farmers to attract funding in return for the biodiversity they support.

Funded by the Federal Innovation Connections Programme.

The Digital twin project leverages Simulation Modelling and advanced analytics to optimise target objectives. It aims to explore new scenarios and assess various business decisions, using a developed simulation-optimisation framework.

As models grow complex, traditional trial and error methods for parameter tuning become inefficient. The proposed system, modelled as a Digital Twin, incorporates live IoT data for intelligent, data-driven analysis.

Utilising reinforcement learning and software like AnyLogic, alongside reviewing metaheuristic optimisation algorithms, the solution aims to optimise Blast Furnace 5 stock house operations at BlueScope steel, focusing on aspects like discharge rates and bin capacities to enhance efficiency and minimise material degradation.

Further, integration with Telstra Digital Twin and other AI/ML services is explored to augment the platform's predictive analytics capabilities, aiming for live, production-ready deployment in the project's second phase.

Informing the development of AIOT-powered rationed care monitoring in Australian residential aged care facilities.

Personalised Travel recommendations and itineraries powered by robust AI. The project's focus on addressing challenges related to input noise, including background noise, accents, and speech recognition errors, reflects a commitment to innovation in the field of AIOT.

Deep learning enabled wearable sensors to facilitate a predictive warning framework aimed at improving the occupational health and safety of steel construction workers.

A project is proposed to explore the integration of Artificial Intelligence and the Internet of Things (AIoT) in satellite systems, addressing key challenges in sustainable development both in space and on Earth. By investigating AI-enhanced satellite-based IoT systems (SIoT), the project aims to improve satellite autonomy, debris management, and orbital traffic control, contributing directly to Space Domain Awareness (SDA) and Space Traffic Management (STM).

AIoT’s role in enabling autonomous decision-making is central to this study, focusing on how satellites can operate independently of constant human intervention to ensure safe and efficient operations in increasingly congested orbital environments. The project also highlights the importance of international collaboration among startups, scholars, and industry stakeholders in developing innovative solutions, particularly in non-military space activities such as debris removal and space traffic management.

Additionally, the project will investigate how AIoT is applied in CubeSats, autonomous robotics, and satellite miniaturization to enhance system performance and efficiency. The aim is to develop resilient satellite systems that improve SDA by enabling real-time tracking and management of space assets while promoting sustainable and safe space operations.

By leveraging AIoT to address both technical and socio-cultural challenges, the project envisions new pathways for innovation, international cooperation, and space-based activities, supporting the broader goals of resilient infrastructure and sustainable industrialization as outlined in the United Nations' Sustainable Development Goals.

The Transformative Tourism Data Management (TMD) platform builds on the previous “Next-Gen VIC: AIOT-Powered Interactive Kiosk” project and aims to establish a comprehensive Tourism AIoT ecosystem. This initiative addresses critical challenges in current content management systems by enhancing data integrity, ensuring real-time data availability, and improving user experiences through personalized interactions. Following are the main components to be establish in this project:
1. Retrieval-Augmented Generation (RAG): RAG techniques will provide highly relevant and personalised content by analyzing user data, preferences, and real-time context. This will deliver tailored travel recommendations, itineraries, and promotional offers. Natural Language Processing (NLP) capabilities will enable users to interact with the system using natural language queries, enhancing accessibility and user satisfaction.
2. iBeacon with Mobile Tourism App: IoT beacons deployed across tourism sites will collect real-time data on visitor movements, environmental conditions, and operational statuses. This data will integrate with a mobile tourism app, providing visitors with real-time, location-based information and notifications to enhance their overall experience. Edge computing will process this data for immediate insights, while more extensive analysis will be conducted in the cloud.
3. Web3 and NFT Integration: Web3 technologies will enable a decentralised, user-centric internet, giving users greater control over their data and interactions. Non-fungible tokens (NFTs) will be used to gamify user engagement, providing unique digital collectibles, rewards, or access to exclusive experiences. This gamification encourages tourists to explore more attractions and engage deeply with local culture.
The TMD project aims to create a sustainable and scalable Tourism IT ecosystem, focusing on infrastructure development, capacity building, research, and operational efficiency. By leveraging RAG, iBeacon with a mobile app, Web3, NFTs, and a robust CMS, the platform will meet the evolving needs of the tourism industry, setting a new standard for technological integration and user engagement.

The project builds on years of fundamental research in 3D human motion analysis at UOW. It aims to develop
a portable and medical-grade device for objective motion measurement and assessment using edge computing,
IoT, and machine learning technologies. This device, equipped with a 3D sensor and an edge computing
system, is designed to capture upper-limb movements of patients. Based on the previous research outcomes, a
set of algorithms that can be deployed to a cloud platform will be developed and validated for objective motion
assessment against standardized criteria like the Action Research Arm Test (ARAT), commonly used in
medical practice. Additionally, the system will leverage recent advances in generative AI to provide natural
language-based explanations of the assessment.
The device will undergo testing in both laboratory settings and with real patient datasets collected by medical
specialists. The project will also explore and analyse the business case and market potential for
commercialization of the device and algorithms.
The expected outcomes of the project include:
• A portable edge-computing device with a 3D sensor and algorithms for capturing and preprocessing
upper-limb motion.
• A suite of machine learning algorithms to analyse the captured motion for objective assessments and
the automatic generation of natural language commentary.
• Both the device and algorithms will be tested in laboratory settings and on real-world datasets.
Testing will be performed on individual algorithms and in a semi-integrated system.
• A report analysing the market opportunities and commercialization pathway for the device and
algorithms.
These outcomes are expected to advance the system's technology readiness level (TRL) from current 2 to a
level between 3 and 4. Along with the business case, the project will be ready to seek further funding from
government, industry, or venture capital, with a particular interest in Australia's Economic Accelerator (AEA)
Ignite scheme.

The Australian government has implemented several urban drinking water quality monitoring programs and frameworks guided by the "Australian Drinking Water Guidelines" (ADWG), emphasizing a multi-barrier approach and real-time monitoring. CSIRO has employed satellite imagery and sensors to monitor algae and sediment plumes and various urban deployments of IoT sensors and processing systems for water quality monitoring. However, no practical system currently integrates distributed AI IoT devices with blockchain, edge-cloud deep learning and multimodal large language models (LLM) for intelligent drinking water quality monitoring. The UrbanPureGuardian project proposes strategically deploying an advanced AI-IoT sensor network across urban water supply system, aiming to establish a framework for drinking water safety prediction, alerting, and emergency response mechanisms based on AI-IoT, blockchain, deep learning algorithms and LLM by leveraging real-time water quality changes, environmental changes, water safety indicators, and emergency plans. These sensors continuously monitor critical water quality parameters, including pH levels, turbidity, chlorine concentration, and conductivity. A real-time data collection platform detects anomalies or deviations indicating potential water quality risks by 5G telecommunication. The core innovation of this project lies in the sophisticated AI-driven analysis of the collected water quality data, edge sensor networking, blockchain data recording, and utilizing an edge-cloud architecture for deep-learning-based model training and deployment. A fine-tuned large language model infers and generates emergency plans, which can be integrated with the pipeline control system to adjust the water network automatically. By solar power, AI IoT devices can autonomously perform data collection, inference, and execution tasks, communicating with the cloud when necessary, thus enhancing the system's environmental adaptability and efficiency while reducing latency and instability caused by network communication. The study cases have already validated the proper functionality of the technological systems and the adequacy of existing facilities, such as the framework of Industry IoT integrating with blockchain, the deep learning algorithm deployed in cloud computing, cybersecurity and traffic prediction. These practical experiences will furnish theoretical underpinnings for the project. The research team will comprehensively investigate methodologies such as mitigating interference and noise in water quality sensor measurements, conducting on-site testing of the developed and calibrated water IoT sensors, and evaluating and identifying barriers and facilitators with industrial partners. The primary output of this seed project will be the UrbanPureGuardian AI IoT-based sensor network system framework and prototype, which will be supported by blockchain, deep-learning algorithm, and multimodal LLM. The project and research are expected to yield valuable publications and system models, and the team will collaborate with industry partners to advance and promote this technology within the sector.

The research questions that this research project aims to address are:
1) Which machine learning algorithms are most effective for forecasting energy consumption at the early stage of building design? and
2) Which machine learning algorithms are most effective for forecasting energy consumption during the preliminary stages of project design?

Therefore, the objective of this research is to develop a newly integrated model that includes three interconnected components (including an energy analysis simulation model for multiple stakeholders such as architects and engineers, a predictive model using modern machine learning algorithms, and an optimization model using new AI algorithms) to predict and optimize energy consumption for various building design options in the early stages of a project.

The expected outcome of this collaborative project is to develop a data-driven framework that aids in making informed decisions during the early design stages, ultimately leading to more energy-efficient buildings. By leveraging AI and ML algorithms, this framework is expected to significantly enhance energy efficiency in building designs, addressing the global challenge of high energy consumption in buildings.