Abstract

The human arm is capable of performing fast targeted movements with high precision, say in pointing with a mouse cursor, but is inherently ‘soft’ due to the muscles, tendons and other tissues of which it is composed. Robot arms are also becoming softer, to enable robustness when operating in real-world environments, and to make them safer to use around people. But softness comes at a price, typically an increase in the complexity of the control required for a given task speed/accuracy requirement. Here we explore how fast and precise joint movements can be simply and effectively performed in a soft robot arm, by taking inspiration from the human arm. First, viscoelastic actuator-tendon systems in an agonist-antagonist setup provide joints with inherent damping, and stiffness that can be varied in real-time through co-contraction. Second, a light-weight and learnable inverse model for each joint enables a fast ballistic phase that drives the arm close to a desired equilibrium point and co-contraction tuple, while the final adjustment is done by a feedback controller. The approach is embodied in the GummiArm, a robot which can almost entirely be printed on hobby-grade 3D printers. This enables rapid and iterative co-exploration of ‘brain’ and ‘body’, and provides a great platform for developing adaptive and bio-inspired behaviours.

DOI

10.1007/978-3-319-43488-9_22

Publication Date

2016-08-10

Event

14th International Conference on Simulation of Adaptive Behavior, SAB 2016

Publication Title

Lecture Notes in Computer Science

Volume

9825

Publisher

Springer

ISBN

9783319434872

ISSN

1611-3349

Embargo Period

2024-11-22

Keywords

Computers

First Page

244

Last Page

255

Additional Files
UoP_Deposit_Agreement v1.1 20160217.pdf (123 kB)
Download

Share

COinS