Title: Self-Supervised Robot Learning

Abstract:

Supervised learning has been used in robotics to solve various tasks like navigation, fine manipulation, etc.  While it has shown a promising result, in most cases the supervision comes from the human agent.  However, relying on human is a huge bottleneck to scale up these approaches.  In this thesis, we try to circumvent human supervision and try to solve the task in a self-supervised manner.  More specifically, we have applied self-supervised learning to 2 tasks, for drone navigation and exploration in reinforcement learning.  In drone navigation, drone first randomly collides with lots of objects and based on it, it learns a policy to avoid them.  In reinforcement learning, we try to learn exploration policy in a self-supervised manner,  for both stochastic and deterministic environment.

Committee:

Abhinav Gupta (advisor)

Martial Hebert

Lerrel Pinto