Research Interests: Explainability, model learning, intrinsic motivation, reinforcement learning, information theory and complex systems

Research Description: As part of the ORCA HUB, I am interested in leveraging new methods for explanation of behaviour in autonomous and semi-autonomous robots. These methods will allow human operators to monitor the actual and future state of activities in a fleet of robots. The methods include activity model learning via program induction, explainability in hierarchical reinforcement learning polices, induction of conditional activity diagrams and planning with predicates. The designed system will allow for teleological and counterfactual explanations, condition monitoring, inspection and online re-planing.