User:S bhuvaneswari

From RISE
Jump to: navigation, search

About - I am a Project Associate at RISE Lab, IIT Madras working under Dr. Balaraman Ravindran. I completed my undergraduate B.Tech degree in Electrical Engineering from IIT Mandi in the year 2014.

Areas of Interest - Robotics, Reinforcement learning

Courses auditing in IIT Madras - Reinforcement Learning offered by Dr. Balaraman Ravindran

Current Project - Learning Robotic Affordances

Affordances of an object are the set of actions/activities that an object affords. For example, a dosa affords eating, a glass affords drinking, a pen writing with, a paper affords writing on etc. Human beings seem to easily learn such affordances from experiment and experience. The aim is to understand and propose a way of learning affordances by a robot using ideas from reinforcement learning. For this we try to look at an object as something that affords many possible 'options'(as proposed by Sutton et al in their paper Between MDPs and Semi-MDPs: A Framework for temporal abstraction in Reinforcement Learning) when in different states and try to learn these options.

Contact - bhuvan.s91@gmail.com

Personal tools
Namespaces
Variants
Actions
Navigation
Toolbox