We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Follow the Object: Curriculum Learning for Manipulation Tasks with Imagined Goals

Abstract: Learning robot manipulation through deep reinforcement learning in environments with sparse rewards is a challenging task. In this paper we address this problem by introducing a notion of imaginary object goals. For a given manipulation task, the object of interest is first trained to reach a desired target position on its own, without being manipulated, through physically realistic simulations. The object policy is then leveraged to build a predictive model of plausible object trajectories providing the robot with a curriculum of incrementally more difficult object goals to reach during training. The proposed algorithm, Follow the Object (FO), has been evaluated on 7 MuJoCo environments requiring increasing degree of exploration, and has achieved higher success rates compared to alternative algorithms. In particularly challenging learning scenarios, e.g. where the object's initial and target positions are far apart, our approach can still learn a policy whereas competing methods currently fail.
Comments: To appear in Deep Reinforcement Learning Workshop, NeurIPS 2021
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Robotics (cs.RO); Machine Learning (stat.ML)
Cite as: arXiv:2008.02066 [cs.LG]
  (or arXiv:2008.02066v2 [cs.LG] for this version)

Submission history

From: Ozsel Kilinc [view email]
[v1] Wed, 5 Aug 2020 12:19:14 GMT (3629kb,D)
[v2] Thu, 11 Nov 2021 07:17:03 GMT (3610kb,D)

Link back to: arXiv, form interface, contact.