My academic interests lie in the fields of reinforcement learning, life-long learning, bio-inspired learning algorithms and optimization applied to control problems. My current research focuses on the safe and sample-efficient training of reinforcement learning agents through the autonomous inference of domain priors, as well as other mechanisms that contribute towards the agent's domain-awareness.
News & Updates
Our paper "Learning Transferable Domain Priors for Safe Exploration in Reinforcement Learning" has been accepted at IJCNN 2020
Localization, mapping, path planning and control
Object recognition using the EvoBot
Multiagent mapping application
EvoBot executing energy optimal navigation