Perception & world models
Trust the world a system perceives.
How can a physical AI system recognise when its understanding of the world has been manipulated?
Ongoing research examines the observations and predictions behind physical decisions. The question is how adversarial inputs can distort an agent's view of its environment or its expectations about what an action will do.
Research areas
The integrity of observations used to understand the environment.
Trust in world models and their predictions about physical actions.
Publications & related resources
Content placeholder
Projects, publications and research materials will be added here.
Contribute to this research.
Explore student research, internships and collaboration with the lab.