Page 31 - Human Inspired Dexterity in Robotic Manipulation

CHAPTER 3

Sensorimotor Learning of

Dexterous Manipulation

Qiushi Fu*, Marco Santello †
*
Mechanical and Aerospace Engineering, University of Central Florida, Orlando, FL, United States
†
School of Biological and Health Systems Engineering, Arizona State University, Tempe, AZ, United States

Contents
3.1 Introduction 27
3.2 Learning Manipulation: Theoretical and Experimental Evidence 29
3.2.1 Background 29
3.2.2 Interaction Between Multiple Sensorimotor Processes Underlies
Learning Dexterous Manipulation 31
3.2.2.1 Subjects and Apparatus 31
3.2.3 Protocols 32
3.2.3.1 Data Analysis 34
3.2.3.2 Model and Simulation 35
3.2.3.3 Results 38
3.2.3.4 Discussion 46
3.3 Lessons Learned From Human Data and Potential Applications
to Robotic Dexterous Manipulation 48
3.4 Conclusions 50
References 50
Further Reading 52

3.1 INTRODUCTION
The human hand is a unique sensorimotor system with unparalleled versa-
tility. Throughout the early stages of motor development and before acquir-
ing the ability to walk, humans learned to explore the environment and its
dynamics by interacting with objects and tools. The hand’s complex biome-
chanical architecture, together with the nervous system’s ability to integrate
sensory feedback with motor commands, dictate the extent to which such
interactions are successfully learned to be then executed in a consistent and
precise fashion (hand anatomy and neural mechanisms are presented in this
chapter and Chapter 2 and reviewed in [1]).

Human Inspired Dexterity in Robotic Manipulation © 2018 Elsevier Inc.
https://doi.org/10.1016/B978-0-12-813385-9.00003-0 All rights reserved. 27