Dr Zito’s main research interest is in mapping perception into actions to model intelligent behaviours for next-gen AI systems. Perception is hard and uncertain in real-world tasks, autonomous systems have the need of computational models for i) inferring the current state of the world from noisy sensors; ii) making predictions on how their actions will effect changes in the state, given unknown or uncertain dynamics; and select the most resonable actions for safety completing the desired task. Dr Zito employs a combination of model-based and data-driven machine learning approaches to provide robust autonomous reasoning to machines that have to interact with the physical world.