Foundation-model-guided world models and predictive control for autonomous remote handling in extreme environments
Explore autonomous robotic manipulation using foundation models to improve decision-making under uncertainty. Engage with cutting-edge predictive control and world model methods for hazardous, unstructured environments.
AI-generated overview
Project Description
Project Overview
This project seeks to develop intelligent robotic systems for autonomous remote handling in extreme and hazardous environments such as fusion and nuclear facilities. It will explore how foundation models can guide world models and predictive control to improve decision-making, robustness, and adaptability when sensing is limited and uncertainty is high.
What You Will Do
The candidate will work with multidisciplinary research teams to combine learning-based prediction, model predictive control, and semantic priors from foundation models. The candidate will develop methods enabling robots to reason about dynamics, task constraints, and safe action selection for complex manipulation tasks.
Expected Outcomes
The research will contribute to next-generation autonomous robotic systems suited for challenging industrial applications, delivering advancements in embodied AI and autonomous robotics with both academic publications and practical implementations.
Why This Matters
Autonomous robotic handling in hazardous environments significantly reduces human risk and improves operational efficiency. By leveraging foundation models, this work aims to enhance robot capabilities in uncertain and unstructured environments, addressing critical challenges in fusion and nuclear sectors.
Entry Requirements
How to Apply
Eligibility
Supervisor Profile
Prof Amir Ghalamzan is an Associate Professor at the University of Sheffield specializing in robotics, robotic manipulation, data-driven control, and foundational models. His research integrates machine learning and control techniques to advance autonomous robotic systems. He collaborates with leading robotics labs and has a solid record of impactful publications.