Learning Transferable Policies for Monocular Reactive MAV Control

TL;DR

This paper introduces a framework for learning transferable motion policies to enable monocular reactive MAV control, demonstrating successful real-world outdoor flight experiments in cluttered environments.

Contribution

It proposes a generic transfer learning framework for autonomous MAV control that leverages data from related source domains to improve performance in target environments.

Findings

Effective transfer of policies in outdoor cluttered environments

Successful real-world MAV flight experiments

Improved adaptability of reactive control policies

Abstract

The ability to transfer knowledge gained in previous tasks into new contexts is one of the most important mechanisms of human learning. Despite this, adapting autonomous behavior to be reused in partially similar settings is still an open problem in current robotics research. In this paper, we take a small step in this direction and propose a generic framework for learning transferable motion policies. Our goal is to solve a learning problem in a target domain by utilizing the training data in a different but related source domain. We present this in the context of an autonomous MAV flight using monocular reactive control, and demonstrate the efficacy of our proposed approach through extensive real-world flight experiments in outdoor cluttered environments.