Introduction
Household waste recycling is a cornerstone of the UK’s environmental strategy, yet major challenges remain.
In 2021, the UK achieved a 44.6% recycling rate for household waste, with significant regional variation. Despite progress, surveys indicate that UK households discard billions of plastic items annually, with only a fraction being recycled domestically. Improving recycling efficiency and accuracy is therefore critical to reducing landfill use and promoting a circular economy.
This project explores how interactive information access (IIA) and AI-powered systems can enhance recycling by enabling accurate, efficient, and sustainable identification and sorting of waste materials. Traditional recycling methods often rely on manual sorting, which is time-consuming and error-prone. Computer vision, multimodal information processing, and interactive retrieval technologies could be used to design systems that distinguish materials, reduce contamination, and improve recycling throughput.
The PhD will focus on developing advanced controllers and learning mechanisms inspired by human motor control.
These will combine a priori internal models with adaptive learning from multimodal data, including image recognition and sensor inputs, to improve robotic grasping and sorting of recyclables. Generative AI may also be explored to enrich multimodal data and increase the robustness of control strategies.
Key research questions include:
How can interactive information access and multimodal learning reduce the error rate in material identification and sorting?
What control strategies allow robots to handle complex, irregular waste objects with poorly defined material properties?
How can sustainability be embedded into system design, ensuring reduced carbon intensity and improved efficiency?
The project will employ experimental robotics, multimodal data modelling, generative AI, and evaluation in collaboration with industrial recycling partners. Outcomes will support more efficient, sustainable, and scalable recycling infrastructures.