Publication

Dynamic cloth manipulation with deep reinforcement learning

Conference Article

Conference

IEEE International Conference on Robotics and Automation (ICRA)

Edition

2020

Pages

4630-4636

Doc link

https://ras.papercept.net/proceedings/ICRA20/1531.pdf

File

Download the digital copy of the doc pdf document

Abstract

In this paper we present a Deep Reinforcement Learning approach to solve dynamic cloth manipulation tasks. Differing from the case of rigid objects, we stress that the followed trajectory (including speed and acceleration) has a decisive influence on the final state of cloth, which can greatly vary even if the positions reached by the grasped points are the same. We explore how goal positions for non-grasped points can be attained through learning adequate trajectories for the grasped points. Our approach uses few demonstrations to improve control policy learning, and a sparse reward approach to avoid engineering complex reward functions. Since perception of textiles is challenging, we also study different state representations to assess the minimum observation space required for learning to succeed. Finally, we compare different combinations of control policy encodings, demonstrations, and sparse reward learning techniques, and show that our proposed approach can learn dynamic cloth manipulation in an efficient way, i.e., using a reduced observation space, a few demonstrations, and a sparse reward.

Categories

intelligent robots, manipulators.

Author keywords

Deep Reinforcement Learning, Dynamic Manipulation, Learning in Simulation, Deformable Object manipulation

Scientific reference

R. Jangir, G. Alenyà and C. Torras. Dynamic cloth manipulation with deep reinforcement learning, 2020 IEEE International Conference on Robotics and Automation, 2020, Paris, France, pp. 4630-4636, to appear.