CPS 300 - Graduate Seminar

Student Presentations

Mobile Robot Navigationwith Neural Maps

Michail G. LagoudakisDepartment of Computer Science

Duke University

(Center for Advanced Computer Studies)

(University of Southwestern Louisiana)

Motivation

Autonomous Robots– Planetary Exploration– Service Robots– Inspection Robots

Biologically-Inspired Robots– Animal Learning and Behavior– Neural Network Models

Talk Outline

Background– Mobile Robot Navigation– Neural Maps

Navigation with Neural Maps The Polar Neural Map Implementation on a Nomad 200 Results Conclusions

Mobile Robots

RWI B14 RWI B21

Nomad 200Nomad 150Nomad XR4000

Mobile Robot Navigation

Global– Map-Based

– Deliberative

– Slow

Local– Sensory-Based

– Reactive

– Fast

Navigation Subproblems

What should I remember?• Cognitive Mapping

Where am I?• Localization

Where should I go?• Path Planning

How can I go?• Motion Control

Animal Navigation

Robot Navigation

Navigation Landscape

Neural Maps

“A localized neural representation of signals in the outer world” [Amari]

Hopfield-type Neural Networks Topologically Ordered Units

Neural Map Property

Amplification through Self-Organization

Neurons

Non-linear Processing Units

Non-Linear Dynamics

ui(t)

W i V (t) i( t)

vi (t 1) (ui(t))dvi (t)

dt(ui(t)) vi(t)

Path Planning with Neural Maps

Network Topology

Path Planning Example 1

Path Planning Example 1

Path Planning Example 1

Path Planning Example 2

Path Planning Example 2

Path Planning Example 2

Problem?

Global Information?

A “Bad” Idea

A “Good” Idea

The Polar Neural Map

Represents the local space. Resembles the distribution

of sensory data. Provides higher resolution

closer to the robot. Conventions:

– Inner Ring: Robot Center

– Outer Ring: Target Direction

Example

Example (...continued)

Example (...continued)

Example (...continued)

Boudreaux (Nomad 200) Nonholonomic Mobile Base Zero Gyro-Radius Max Speeds: 24 in/sec, 60 deg/sec Diameter: 21 in, Height: 31 in Pentium-Based Master PC Linux Operating System Full Wireless 1.6 Mbps Ethernet 16 Sonar Ring (6 in - 255 in) 20 Bump Sensors

System Architecture

Results

Results (…continued)

Results (…continued)

Results (…continued)

Results (…continued)

Results (…continued)

Robot Movie

Enjoy some robotic video footage!

Contributions

Neural Maps for Fast Path Planning The Polar Neural Map Implementation on a Real Robot A Complete Local Navigation Scheme

Future Work

On-board Code Execution Polar and Logarithmic Map Global Navigation Neural Map Self-Organization

More Information

M.Sc. Thesis, Poster– http://www.cs.duke.edu/~mgl/acadpape.html

IEEE ICRA ’99 Paper– Email: mgl@cs.duke.edu

Thank You

TheThe EndEnd!

©1998 - CPS 300 Inc.

Special thanks to

The Robotics and Automation Lab at USLProf. Anthony S. Maida

Prof. Kimon P. ValavanisProf. Bill Z. Manaris