Current Projects

The Vax Project (2024--now):

This project aims to (1) improve vaccine manufacturing processes for rapid development/scale-up, faster, simpler, and more cost-effective manufacturing, (2) develop novel vaccine manufacturing technologies (e.g., standardised antigen presentation, multivalent vaccine) to simplify manufacturing and elevate manufacturing flexibility, and (3) develop economic modelling and simulation modelling tools to inform manufacturing and supply chain decisions, and establish investment cases. Our work here falls into (3) focusing on the performance enhancement of vaccine supply chains.

The CORES Project (2022--Now):

This project is regarding Collective Risk Learning for Supply Chain Disruption Preparedness. Supply chain resilience is a long studied topic of significant impact to our society. As organisations outsource production to one another they create economies-of-scale and reduce prices but also increase risk of disruption cascades if any member of the chain fails. Typically, organisations act alone, rather than as a collective, when predicting risk and deciding on safety inventories. However, risk data an individual organisation can collect and analyse is small, imbalanced, and partial entirely to its own view. When uncertainties increase, this individualistic approach results in chaotic oscillations between stock inflation and stock-outs. Numerous studies proved that increased data sharing and collective decision making would increase resilience, but this has not been plausible as members of the chain fear that information such as capacity and excess stock can be “inferred” by clients, and used opportunistically for cost reduction. Federated learning is an emerging approach in Artificial Intelligence that may help supply chain members collectively optimise resilience, whilst keeping their data private. The approach enables organisations to collaboratively develop a shared prediction model. Here, if one organisation is able to predict risk, its knowledge can be shared, preventing others from stock outs. As the approach can be automated, costs of manual orchestration are avoided. However, FL has not been developed for industrial applications in mind, and research is needed to create suitable algorithms. In this project we will develop and compare suitable FL approaches specifically for risk prediction and collective learning in supply chains, with real use cases in the aerospace manufacturing.

Previous Projects

The Traffic Prediction in Urban Transport Networks Project (2018--2022):

It aims to solve traffic problems, such as traffic congestion, traffic accidents, and long travel times, by predicting traffic states in advance on large-scale road networks to contribute to the performance improvement of Intelligent Transportation Systems (ITSs). In this research project, we collected real-world data from highways around Heathrow Airport in London and from both Los Angeles and Seattle in the USA, analyzed specific traffic patterns on linear roadways, intersections, and the entire road networks, and built novel deep learning models on these different types of roadways for traffic prediction.

The HAR Project (2020--2021):

This project aims to recognise human activities such as jumping, walking, sitting, drinking, and eating various food by analysing and distinguishing patterns among those activities.

The Fall Detection Project (2018):

This project aims to detect falls from various daily activities for the elderly to escape from dangerous situations when they fall. In this project, we

  • Build a sensor board embedded with an accelerometer and gyroscope to collect data.
  • Develop machine learning algorithms for detecting falls from daily activities.
  • Evaluate the proposed algorithms by conducting experiments on the collected real-world datasets.