Multi-Fingered In-Hand Manipulation with Various Object Properties Using Graph Convolutional Networks and Distributed Tactile Sensors

TL;DR

This paper introduces a graph convolutional network-based control method for multi-fingered robotic hands, enabling stable in-hand manipulation of various objects using tactile sensors and object property labels.

Contribution

The novel use of GCNs to process complex tactile data and incorporate object properties for improved dexterous manipulation in robotic hands.

Findings

High success rates in in-hand manipulation tasks.

Fragile objects deformed less with object property labels.

GCN learned geodesical features from tactile data.

Abstract

Multi-fingered hands could be used to achieve many dexterous manipulation tasks, similarly to humans, and tactile sensing could enhance the manipulation stability for a variety of objects. However, tactile sensors on multi-fingered hands have a variety of sizes and shapes. Convolutional neural networks (CNN) can be useful for processing tactile information, but the information from multi-fingered hands needs an arbitrary pre-processing, as CNNs require a rectangularly shaped input, which may lead to unstable results. Therefore, how to process such complex shaped tactile information and utilize it for achieving manipulation skills is still an open issue. This paper presents a control method based on a graph convolutional network (GCN) which extracts geodesical features from the tactile data with complicated sensor alignments. Moreover, object property labels are provided to the GCN to adjust in-hand manipulation motions. Distributed tri-axial tactile sensors are mounted on the fingertips, finger phalanges and palm of an Allegro hand, resulting in 1152 tactile measurements. Training data is collected with a data-glove to transfer human dexterous manipulation directly to the robot hand. The GCN achieved high success rates for in-hand manipulation. We also confirmed that fragile objects were deformed less when correct object labels were provided to the GCN. When visualizing the activation of the GCN with a PCA, we verified that the network acquired geodesical features. Our method achieved stable manipulation even when an experimenter pulled a grasped object and for untrained objects.