Dynamic Manipulation of Flexible Objects with Torque Sequence Using a Deep Neural Network

Kawaharazuka, Kento; Ogawa, Teiichiro; Tamura, Juntaro; Nabeshima, Cota

doi:10.48550/arxiv.1901.10142

Cited by 1 publication

(1 citation statement)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Alternative approaches that employ trial-and-error-based machine learning methods ( [15], [16], [17], [18], [19], [20]) to map observations directly to actions have shown interesting results on robotic manipulation of deformable objects, but they restrict themselves to static manipulation as well. Regarding dynamic manipulation, [21] developed a control method to realize a target state by calculating an optimized time-series joint torque command, but they mainly consider only a 2D actuation that allows only simple manipulation tasks and [22] considers only quasi-static movements.…”

Section: Introductionmentioning

confidence: 99%

Dynamic Cloth Manipulation with Deep Reinforcement Learning

Jangir¹,

Alenyà²,

Torras³

2019

Preprint

View full text Add to dashboard Cite

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.

show abstract