Sample efficient learning of path following and obstacle avoidance behavior for Quadrotors

Stefan Stevsic, Tobias Nägeli, Javier Alonso Mora, Otmar Hilliges

Research output: Contribution to journalArticleScientificpeer-review

10 Citations (Scopus)
48 Downloads (Pure)

Abstract

In this letter, we propose an algorithm for the training of neural network control policies for quadrotors. The learned control policy computes control commands directly from sensor inputs and is, hence, computationally efficient. An imitation learning algorithm produces a policy that reproduces the behavior of a supervisor. The supervisor provides demonstrations of path following and collision avoidance maneuvers. Due to the generalization ability of neural networks, the resulting policy performs local collision avoidance, while following a global reference path. The algorithm uses a time-free model-predictive path-following controller as a supervisor. The controller generates demonstrations by following few example paths. This enables an easy-to-implement learning algorithm that is robust to errors of the model used in the model-predictive controller. The policy is trained on the real quadrotor, which requires collision-free exploration around the example path. An adapted version of the supervisor is used to enable exploration. Thus, the policy can be trained from a relatively small number of examples on the real quadrotor, making the training sample efficient
Original languageEnglish
Pages (from-to)3852-3859
JournalIEEE Robotics and Automation Letters
Volume3
Issue number4
DOIs
Publication statusPublished - 2018

Keywords

  • Collision avoidance
  • deep learning in robotics and automation

Fingerprint Dive into the research topics of 'Sample efficient learning of path following and obstacle avoidance behavior for Quadrotors'. Together they form a unique fingerprint.

Cite this