Learning to push by grasping: Using multiple tasks for effective learning

Lerrel Pinto, Abhinav Gupta

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Recently, end-to-end learning frameworks are gaining prevalence in the field of robot control. These frameworks input states/images and directly predict the torques or the action parameters. However, these approaches are often critiqued due to their huge data requirements for learning a task. The argument of the difficulty in scalability to multiple tasks is well founded, since training these tasks often require hundreds or thousands of examples. But do end-to-end approaches need to learn a unique model for every task? Intuitively, it seems that sharing across tasks should help since all tasks require some common understanding of the environment. In this paper, we attempt to take the next step in data-driven end-to-end learning frameworks: move from the realm of task-specific models to joint learning of multiple robot tasks. In an astonishing result we show that models with multi-task learning tend to perform better than task-specific models trained with same amounts of data. For example, a deep-network learned with 2.5K grasp and 2.5K push examples performs better on grasping than a network trained on 5K grasp examples.

Original languageEnglish (US)
Title of host publicationICRA 2017 - IEEE International Conference on Robotics and Automation
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages2161-2168
Number of pages8
ISBN (Electronic)9781509046331
DOIs
StatePublished - Jul 21 2017
Event2017 IEEE International Conference on Robotics and Automation, ICRA 2017 - Singapore, Singapore
Duration: May 29 2017Jun 3 2017

Publication series

NameProceedings - IEEE International Conference on Robotics and Automation
ISSN (Print)1050-4729

Conference

Conference2017 IEEE International Conference on Robotics and Automation, ICRA 2017
CountrySingapore
CitySingapore
Period5/29/176/3/17

ASJC Scopus subject areas

  • Software
  • Control and Systems Engineering
  • Artificial Intelligence
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Learning to push by grasping: Using multiple tasks for effective learning'. Together they form a unique fingerprint.

Cite this