Connectionist Robot Motion Planning

Connectionist Robot Motion Planning: A Neurally-Inspired Approach to Visually-Guided Reaching is the third series in a cluster of books on robotics and related areas as part of the Perspectives in Artificial Intelligence Series. This series focuses on an experimental paradigm using the MURPHY system to tackle critical issues surrounding robot motion planning. MURPHY is a robot-camera system developed to explore an approach to the kinematics of sensory-motor learning and control for a multi-link arm. Organized into eight chapters, this book describes the guiding of a multi-link arm to visual targets in a cluttered workspace. It primarily focuses on “ecological” solutions that are relevant to the typical visually guided reaching behaviors of humans and animals in natural environments. Algorithms that work well in unmodeled workspaces whose effective layouts can change from moment to moment with movements of the eyes, head, limbs, and body are also presented. This book also examines the strengths of neurally inspired connectionist representations and the utility of heuristic search when good performance, even if suboptimal, is adequate for the task. The co-evolution of MURPHY’s design with the brain, presumably in response to similar computational pressures, is described in the concluding chapters, specifically presenting the division of labor between programmed-feedforward and visual-feedback modes of limb control. Design engineers in the fields of biology, neurophysiology, and cognitive psychology will find this book of great value.

Preface

Acknowledgments

1 Introduction

2 MURPHY's Organization

The Physical Setup

The Connectionist Architecture

Representational Choices

MURPHY's Kinematics

Traditional Methods

Connectionist Kinematics

The Sequential Controller

3 How MURPHY Learns

Learning by Doing

Abridged History of the Idea

Motivating Sigma-Pi Learning

Connectionist Supervised Associative Learning

A Historical Quandary

New Algorithms Are More Powerful, Less Biological

Changing Assumptions

Learning Functions with Lookup-Tables

Building Receptive Fields with Multiplication

Sigma-Pi Learning

The Sigma-Pi Unit

The Learning Rule

An Example

Generalization to Novel Inputs

K-d Tree Reimplementation

4 MURPHY in Action

Planning with Gradient Descent

Motion Planning with Obstacles

Four Visual Routines

The Search Procedure

Discussion

Visual-Feedback Control

System Performance and Scaling Behavior

Implementation and Performance Notes

Scaling MURPHY Up

5 Robotics Issues

Learning vs. Built-in Models

Style of Representation

Using the Full Visual Channel

Styles of Robot Motion Planning

Artificial Potential Fields: Local Methods

Geometric Motion Planning: Global Methods

Hybrid Methods: Incorporating Heuristic Search

MURPHY Uses Direct Heuristic Search

6 Psychological Issues

Psychoanalyzing MURPHY

Development and Plasticity in Limb Control

The Necessity of Active Sensory-Motor Learning

Developmental Evidence for Two Modes of Control

Possible Evidence for Local Learning

Models, Imagery, Practice, and Stability

Building and Using Mental Models

Mental Imagery

Mental Practice: Learning by Thinking

Corollary Discharges and Perceptual Stability

Summary of Psychological Issues

7 Biological Issues

The Muscle Interface

A Cortical Basis for Visual Limb Control

A Programmed-Feedforward Neural Subsystem

A Visual-Feedback Neural Subsystem

Summary of Biological Issues

8 Conclusions

Scientific and Engineering Lessons Learned

Learning by Doing

Learning with Lookup-Tables

The Connectionist Architecture

Mental Models, Heuristic Search

Reflections on Brain and Behavior

Future Directions

Pragmatic Extensions

Theoretical Extensions

Bibliography

Index