Inverse Reinforcement Learning

Humans perform several dynamic tasks with ease by learning a high-level decision-making policy. Whenever a new scenario is encountered decisions are made based on this policy rather than complete re-learning. One of the key goals of this research is to enable robots to learn from humans in a way that is intuitive and natural so that they can be more easily integrated into a wide range of tasks and environments.