Download Full Version ebookmass - Visit ebookmass.

com

Learning Control Zhang

https://ebookmass.com/product/learning-control-zhang/

OR CLICK HERE

DOWLOAD NOW

Discover More Ebook - Explore Now at ebookmass.com

 Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Learning Control Dan Zhang

https://ebookmass.com/product/learning-control-dan-zhang/

ebookmass.com

MOTOR LEARNING AND CONTROL Concepts and Applications

Richard A. Magill

https://ebookmass.com/product/motor-learning-and-control-concepts-and-
applications-richard-a-magill/

ebookmass.com

Craving Control Faust

https://ebookmass.com/product/craving-control-faust/

ebookmass.com

Communicating Across Cultures and Languages in the Health

Care Setting: Voices of Care 1st Edition Claire Penn

https://ebookmass.com/product/communicating-across-cultures-and-
languages-in-the-health-care-setting-voices-of-care-1st-edition-
claire-penn/
ebookmass.com
PMP Project Management Professional All-in-One Exam Guide
Joseph Phillips

https://ebookmass.com/product/pmp-project-management-professional-all-
in-one-exam-guide-joseph-phillips/

ebookmass.com

Climate Economics: A Call for More Pluralism And

Responsibility Michael Roos

https://ebookmass.com/product/climate-economics-a-call-for-more-
pluralism-and-responsibility-michael-roos/

ebookmass.com

Code Name: Omega (Jameson Force Security Book 10) Sawyer

Bennett

https://ebookmass.com/product/code-name-omega-jameson-force-security-
book-10-sawyer-bennett/

ebookmass.com

Oxford Studies in Political Philosophy, Volume 4 David

Sobel

https://ebookmass.com/product/oxford-studies-in-political-philosophy-
volume-4-david-sobel/

ebookmass.com

An Introduction to Policing 9th Edition, (Ebook PDF)

https://ebookmass.com/product/an-introduction-to-policing-9th-edition-
ebook-pdf/

ebookmass.com
Narratives of Difference in an Age of Austerity 1st
Edition Irene Gedalof (Auth.)

https://ebookmass.com/product/narratives-of-difference-in-an-age-of-
austerity-1st-edition-irene-gedalof-auth/

ebookmass.com
Learning Control

Applications in Robotics and

Complex Dynamical Systems

FIRST EDITION

Dan Zhang
York University, Toronto, ON, Canada

Bin Wei
Algoma University, Sault Ste Marie, ON, Canada
Table of Contents

Cover image

Title page

Copyright

List of contributors

Chapter 1: A high-level design process for neural-network controls

through a framework of human personalities

Abstract

1.1. Introduction

1.2. Background

1.3. Proposed methods

1.4. Results

1.5. Conclusions

Appendix 1.A.

References
Chapter 2: Cognitive load estimation for adaptive human–machine
system automation

Abstract

2.1. Introduction

2.2. Noninvasive metrics of cognitive load

2.3. Details of open-loop experiments

2.4. Conclusions and discussions

2.5. List of abbreviations

References

Chapter 3: Comprehensive error analysis beyond system innovations

in Kalman filtering

Abstract

3.1. Introduction

3.2. Standard formulation of Kalman filter after minimum

variance principle

3.3. Alternate formulations of Kalman filter after least squares

principle

3.4. Redundancy contribution in Kalman filtering

3.5. Variance of unit weight and variance component estimation

3.6. Test statistics

 3.7. Real data analysis with multi-sensor integrated kinematic
positioning and navigation

3.8. Remarks

References

Chapter 4: Nonlinear control

Abstract

4.1. System modeling

4.2. Nonlinear control

4.3. Summary

References

Chapter 5: Deep learning approaches in face analysis

Abstract

5.1. Introduction

5.2. Face detection

5.3. Pre-processing steps

5.4. Facial attribute estimation

5.5. Facial expression recognition

5.6. Face recognition

5.7. Discussion and conclusion

 References

Chapter 6: Finite multi-dimensional generalized Gamma Mixture

Model Learning for feature selection

Abstract

6.1. Introduction

6.2. The proposed model

6.3. Parameter estimation

6.4. Model selection using the minimum message length

criterion

6.5. Experimental results

6.6. Conclusion

References

Chapter 7: Variational learning of finite shifted scaled Dirichlet

mixture models

Abstract

7.1. Introduction

7.2. Model specification

7.3. Variational Bayesian learning

7.4. Experimental result

7.5. Conclusion
 Appendix 7.A.

Appendix 7.B.

References

Chapter 8: From traditional to deep learning: Fault diagnosis for

autonomous vehicles

Abstract

8.1. Introduction

8.2. Traditional fault diagnosis

8.3. Deep learning for fault diagnosis

8.4. An example using deep learning for fault detection

8.5. Conclusion

References

Chapter 9: Controlling satellites with reaction wheels

Abstract

9.1. Introduction

9.2. Spacecraft attitude mathematical model

9.3. Attitude tracking

9.4. Actuator dynamics

9.5. Attitude control law

 9.6. Performance analysis

9.7. Conclusions

References

Chapter 10: Vision dynamics-based learning control

Abstract

10.1. Introduction

10.2. Problem definition

10.3. Experiments

10.4. Conclusions

References

Index
Copyright
Elsevier
Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United
Kingdom
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States

Copyright © 2021 Elsevier Inc. All rights reserved.

MATLAB® is a trademark of The MathWorks, Inc. and is used with

permission.
The MathWorks does not warrant the accuracy of the text or
exercises in this book.
This book's use or discussion of MATLAB® software or related
products does not constitute endorsement or sponsorship by The
MathWorks of a particular pedagogical approach or particular use of
the MATLAB® software.

No part of this publication may be reproduced or transmitted in any

form or by any means, electronic or mechanical, including
photocopying, recording, or any information storage and retrieval
system, without permission in writing from the publisher. Details on
how to seek permission, further information about the Publisher's
permissions policies and our arrangements with organizations such
as the Copyright Clearance Center and the Copyright Licensing
Agency, can be found at our website:
www.elsevier.com/permissions.
This book and the individual contributions contained in it are
protected under copyright by the Publisher (other than as may be
noted herein).

Notices
Knowledge and best practice in this field are constantly changing.
As new research and experience broaden our understanding,
changes in research methods, professional practices, or medical
treatment may become necessary.

Practitioners and researchers must always rely on their own

experience and knowledge in evaluating and using any
information, methods, compounds, or experiments described
herein. In using such information or methods they should be
mindful of their own safety and the safety of others, including
parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the
authors, contributors, or editors, assume any liability for any injury
and/or damage to persons or property as a matter of products
liability, negligence or otherwise, or from any use or operation of
any methods, products, instructions, or ideas contained in the
material herein.

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of
Congress

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

ISBN: 978-0-12-822314-7

For information on all Elsevier publications visit our website at

https://www.elsevier.com/books-and-journals
Publisher: Matthew Deans
Acquisitions Editor: Dennis McGonagle
Editorial Project Manager: John Leonard
Production Project Manager: Poulouse Joseph
Designer: Victoria Pearson

Typeset by VTeX
List of contributors
Simge Akay Computer Engineering Department, Bahcesehir
University, Istanbul, Turkey
Basim Alghabashi Concordia Institute for Information Systems
Engineering (CIISE), Concordia University, Montreal, QC, Canada
Mohamed Al Mashrgy Electrical and Computer Engineering
(ECE), Al-Mergib University, Alkhums, Libya
Nafiz Arica Computer Engineering Department, Bahcesehir
University, Istanbul, Turkey
Zeinab Arjmandiasl Concordia Institute for Information Systems
Engineering (CIISE), Concordia University, Montreal, QC, Canada
Muhammad Azam Electrical and Computer Engineering (ECE),
Concordia University, Montreal, QC, Canada
B. Balasingam Department of Electrical and Computer
Engineering, University of Windsor, Windsor, ON, Canada
Jamal Bentahar Concordia Institute for Information Systems
Engineering (CIISE), Concordia University, Montreal, QC, Canada
F. Biondi Faculty of Human Kinetics, University of Windsor,
Windsor, ON, Canada
Aaron Boda Department of Earth and Space Science and
Engineering, Lassonde School of Engineering, York University,
Toronto, ON, Canada
Nizar Bouguila Concordia Institute for Information Systems
Engineering (CIISE), Concordia University, Montreal, QC, Canada
Duygu Cakir Computer Engineering Department, Bahcesehir
University, Istanbul, Turkey
Mark Green Faculty of Science, Ontario Tech University, Oshawa,
ON, Canada
Sorin Grigorescu
Robotics, Vision and Control (ROVIS), Transilvania University of
Brasov, Brasov, Romania
Artificial Intelligence, Elektrobit Automotive, Brasov, Romania
Baoxin Hu Department of Earth and Space Science and
Engineering, Lassonde School of Engineering, York University,
Toronto, ON, Canada
Xishi Huang RS Opto Tech Ltd., Suzhou, Jiangsu, China
M. Khalghollah Schulich School of Engineering, University of
Calgary, Calgary, AB, Canada
Howard Li Department of Electrical and Computer Engineering,
University of New Brunswick, Fredericton, NB, Canada
C.J.B. Macnab Schulich School of Engineering, University of
Calgary, Calgary, AB, Canada
Narges Manouchehri Concordia Institute for Information Systems
Engineering (CIISE), Concordia University, Montreal, QC, Canada
Jun Meng Zhejiang University Robotics Institute, Hangzhou,
China
Afshin Rahimi Department of Mechanical, Automotive and
Materials Engineering, University of Windsor, Windsor, ON, Canada
P. Ramakrishnan Department of Electrical and Computer
Engineering, University of Windsor, Windsor, ON, Canada
Jing Ren Department of Electrical and Computer Engineering,
Ontario Tech University, Oshawa, ON, Canada
Jianguo Wang Department of Earth and Space Science and
Engineering, Lassonde School of Engineering, York University,
Toronto, ON, Canada
Jin Wang Zhejiang University Robotics Institute, Hangzhou,
China
Chapter 1: A high-level design
process for neural-network
controls through a framework of
human personalities
M. Khalghollah; C.J.B. Macnab Schulich School of Engineering,
University of Calgary, Calgary, AB, Canada

Abstract
Current learning systems in the field of feedback control deal
with both unbiased nonlinearities and biased nonlinearities
(where bias is measured at the origin) quite differently.
Unbiased nonlinearities lend themselves to direct adaptive
control methods. Biased systems, on the other hand, typically
require actual learning of the bias term in order to achieve
acceptable error and effort. This paper attempts to unify these
approaches, and to learn to compensate for both types of
nonlinearities simultaneously. To do so we utilize a graphical,
quantitative theory of human personalities, which assumes that
their personalities indicate how people interact with the world
around them using feedback. This biologically-inspired
approach allows us to develop a formal design framework for
tackling this problem. Simulations with a two-link robotic
manipulator demonstrate the utility of the learning design
method, where gravity provides the main biased nonlinearities,
while friction, centripetal, and Coriolis forces are treated as
 unbiased nonlinearities; our neural-network update laws learn
all these robot nonlinearities at the same time. Lyapunov
methods result in stability guaranties for the proposed method.

Keywords
learning control; direct adaptive control; linear quadratic regulator;
gravity compensation; trajectory tracking; robotic manipulators;
personality theory
Chapter Outline

1.1 Introduction
1.2 Background
1.2.1 The CMAC associative-memory neural
network
1.2.2 Unbiased nonlinearities
1.2.3 Direct adaptive control in the presence
of bias
1.2.4 A graphical model of personalities
1.2.5 A computer model of personalities
1.3 Proposed methods
1.3.1 Proposed learning law
1.3.2 Cost functional for optimization
1.3.3 Stability analysis
1.4 Results
1.4.1 Developing a design procedure
1.4.2 Two-link robotic manipulator
1.5 Conclusions
References

1.1 Introduction
This work addresses the difficulty in designing a learning control for
system dynamics that contain both a large bias and significant
nonlinearities, using the example of a robotic manipulator to
develop and test the ideas. The vast majority of learning-control
techniques in the literature addressing tracking control of the robot's
end-effector try to cancel the force of gravity, or compensate for
velocity-dependent nonlinearities that tend to pull the tip off-track,
i.e. friction, Coriolis, and centripetal (FCC) terms. The ad hoc
methods found in the literature that proposed learning both gravity
and FCC terms end up doing so sequentially, not simultaneously, to
the best of our knowledge. In this paper, we suggest the basic
difficulty in trying to achieve both at the same time stems from the
need to approach these two problems differently, i.e. they require
two different types of learning.
Gravity acting on a robot qualifies as a nonlinear term with
significant bias; for a simpler example of a biased nonlinearity,
consider a cosine term near the origin. Control systems generally
endeavor to cancel biased nonlinearities with a feedforward term, i.e.
an open-loop, bias-compensation, or set-point control. A sine
function, on the other hand, provides a simple example of a
nonlinear function with a value of zero at the origin; compensating
for this type of unbiased nonlinearity requires nonlinear feedback. A
learning control for this problem essentially ends up achieving a
memorized feedback term. Since the FCC terms in a robot manipulator
go to zero when the velocity goes to zero, and the need for high-
speed precision remains relatively rare in industrial applications, as
a practical matter the trajectory-tracking control problem without
gravity ends up more like learning sine than cosine. One important
difference from the designer's point of view lies in the choice of
robust weight update method, i.e. designing updates for the weights
(adaptive parameters) that remain robust to disturbances in the
sense of limiting weight drift (overlearning) and preventing bursting
(a sudden increase in error after a period of convergence). For
unbiased nonlinearities, a leakage term forgetting factor that tries to
drive the weights toward zero works well in practice. However, such
a term would directly result in a steady-state error for biased
nonlinearities; for biased nonlinearities the field of adaptive control
offers deadzone, parameter projection, and supervised leakage as
choices for the designer, all of which require some a priori
knowledge of the system's parameters and/or expected disturbances;
as a result many do not classify such methods as true learning
systems.
 Here we use our own previously-proposed graphical model of
human personalities to examine the problem [1]. A resulting
computational model, based on feedback theory, allows a prediction
of the probability distribution of Myers–Briggs behavioral technique
types in the human population, for the J/P and T/F pairs [2] as well
as for the I/E and N/S pairs [3]. This biologically-inspired perspective
allows more intuitive design methods for the high-level thinking
required in today's advanced control-system applications [4]. The
proposed framework provides insight into the nature of the two
types of learning problems outlined above, and it suggests how to
achieve them simultaneously on a robot arm, i.e. like humans do.
Building on the idea of LQR control, the approach results in an
optimization method for the design of all the control parameters
(feedback gains and adaptation rates) in a learning controller—the
first such formal design method appearing in the literature that
extends the LQR approach to nonlinear systems, to our knowledge.
Robotic-manipulator dynamics contain significant nonlinearities;
proposed control methods based on linear-system theory must, at
the very least, assume implementation of a gravity-compensation
term [5]. Actually learning the force of gravity should provide many
advantages, including decreasing engineering-design costs and
achieving real-time payload adaptation. An iterative learning control
[6] has some advantages, but does not address robustness to
disturbances and/or payloads. Radial basis function networks
(RBFNs) can learn to estimate the gravity term in some robot
manipulators [7,8] and some biped robots [9], but the method does
not extend well to multi-link robots due to the curse of dimensionality
when trying to add more inputs into an RBFN network.
The method in [10] learns both gravity and other nonlinearities,
but the very-small leakage term used in order to avoid sag-due-to-
gravity appears insufficient to deal with realistic disturbances. The
authors of [11] proposed an RBFN method that adapts to both
gravity (biased nonlinearity) and FCC terms (unbiased
nonlinearities), but requires knowledge of the inertia matrix—which
implies the designer would, in fact, know the gravity term. In [12],
only gravity compensation gets proposed in the first step of
 y g y p g p p p
backstepping for a flexible-joint robot, and not FCC terms. The same
authors tackle learning all nonlinearities for a Baxter robot in [13],
but the tracking performance requires a set of weights identified
during a learning stage—in practice it would seem the learning stage
would find the large biased nonlinearity and the tracking stage
would fine-tune the performance by compensating for the unbiased
nonlinearities.
In previous work, our research group presented a near-optimal
control [14], which developed an approach to achieve a near-optimal
control signal in the presence of gravity. A cerebellar model
articulation controller (CMAC) [15], with advantages of a fast
adaptation rate and real-time computational ability, was found to
have unique properties for tackling this problem; freezing a set of
supervisory network's local weights when the bias becomes
identified can compensate for the gravity bias term, and further
learning (using leakage) could then fine-tune for the FCC terms. The
disadvantage was the procedure was ad hoc, based on intuitive
insights into the workings of the CMAC.
Developing a formal design method, on the other hand, may have
some far-reaching implications for the field of control. Some have
playfully accused the field of control-systems of having a dirty secret:
designers often choose gains and/or parameters by trial-and-error,
rules-of-thumb, and/or experience. Such methods can be inadequate
when facing contemporary problems of interest, such as robots
interacting with unknown, unstructured environments.
Optimization methods promise a path to solve this problem, but
standard higher-level cognitive frameworks for the design of cost-
functionals remain an open problem to our knowledge. Even in the
case of the linear quadratic regulator (LQR), current theory does not
provide the designer with a method for choosing the Q and R
weighting matrices. For more advanced nonlinear systems that
interact with an environment, researchers struggle with even
creating a suitable cost-functional at the moment. This work
provides a biologically-inspired method for designing cost-
functionals and the value of the weightings. We use a model of
human personalities; choosing a personality directly results in a
 p g p y y
choice of numerical weightings. Thus, a control system designer can
use their intuitive understanding of human personalities at the high-
level design stage. Not only can this avoid trial and error in picking
parameters, but it may also significantly reduce the total number of
parameters needed compared to manual designs. Consider an
analogy to how fuzzy logic proved quite a time-saving invention for
control design, as a result of allowing human intuition to guide the
design of computational decision-making and reducing the number
of parameters that the designer needs to choose; fuzzy control ended
up significantly broadening the field of computer automation, since
many more problems would lend themselves to a cost-effective
and/or time-efficient design solution.
This chapter first gives a background on CMAC neural network
and direct adaptive control methods, as applied to both biased and
unbiased nonlinearities. The Background section ends with a short
introduction to our personality theory, and describes how the
framework enables design of PID+bias controls using a nonlinear
quadratic regulator. In the Proposed Methods section, we show how
to extend the approach to designing an adaptive learning control.
The Results section illustrates a formal engineering design
procedure, based on constraints and objectives, for a simple mass-
spring-mass and then a two-link robotic manipulator.

1.2 Background
1.2.1 The CMAC associative-memory neural
network
This work assumes an associative memory will serve as the
nonlinear approximator without loss of generality, where a weighted
sum of basis functions gives the estimate of nonlinear function
as
 (1.1)

where contains the n inputs, the weights reside in column

vector , and denotes a row vector of basis functions.
Typically one uses Gaussians as the basis functions in a radial basis
function network (RBFN). Since the RBFN's curse of dimensionality
makes it difficult to use for more than three inputs, the learning
controls in this paper use a Cerebellar Model Articulation Controller
(CMAC) instead. The CMAC consists of hypercube-domain basis
functions, or cells, constructed with rectangles, triangles, or splines.
The local nature of the CMAC cells ensures only a small number of
activated cells require calculations, along with their corresponding
weights, during any one time-step of the control.
One could build a CMAC using m offset layers of n-dimensional
arrays. The input activates (indexes) one cell per layer. The
construction of the basis functions in a spline-CMAC follows

(1.2)
where denotes the normalized position within the
cell on the kth input; normalization produces

(1.3)
An example of a spline CMAC with normalized basis functions in
one dimension appears in Fig. 1.1: for two dimensions see Fig. 1.2.
Rather than an impractical allocation of memory for n-dimensional
arrays, the CMAC uses hash-coding to map virtual cells to a one-
dimensional physical memory, since relatively few cells ever get
activated in high-dimensional space during trajectory tracking of a
robot [16] (hash collisions become possible, however unlikely).

FIGURE 1.1 One-input CMAC with spline membership

functions: m = 3 layers and Q = 10 discretizations. The
output is a weighted sum of basis functions.
 FIGURE 1.2 Normalized-spline, two-input CMAC: m = 4 and
Q = 4. The word layers for a CMAC refers to hash-coded
arrays indexed in parallel.

Being a universal nonlinear approximator, the CMAC can

approximate nonlinear function as

(1.4)

where defines the approximation error in the uniform

approximation region : specifically for a
positive constant .

1.2.2 Unbiased nonlinearities

In this work the learning control strategies come from mathematical
techniques developed in the field of direct adaptive control.
Consider a physical system with position x and control input u

(1.5)

where the unbiased nonlinearity has , with constant

parameters , , and . Given desired trajectory ,
then defining state errors , and auxiliary error
leads to an expression of the dynamics suitable for
control-system design

(1.6)

Although a desired trajectory may introduce a system bias, such that

, a practical neural-adaptive control ignores this typically-
small effect and a common design uses leakage [17] multiplied by
positive parameter ν in the robust weight update

(1.7)

(1.8)
where indicates the ideal-weight estimates, positive parameter ν
determines the influence of the robust leakage modification, which
limits weight drift by trying to force the weights to zero. As long as
remains relatively small, the leakage term does not significantly
reduce performance. The definition of z implies positive constant
equates to a derivative gain from PID control—giving an effective
proportional gain . We denote the positive-constant
adaptation gain as because in subsequent sections will we point
out similarities of the neural network to an integral term, i.e. this
work treats a neural network trained with adaptive-control update
laws as just a memorized integral.
Note that the system control actually occurs as a multi-rate signal,
with the feedback running fast enough to approximate a continuous
time signal and the neural network updated at a discrete rate, i.e. a
zero-order-hold discretized signal describes its output characteristics.
Designers often choose leakage for a robust update, and in discrete
time with sampling period Δt the delta-rule update becomes

(1.9)

Note the delta-rule simply results in the neural network estimating

the discrete-time model rather than the continuous-time model using
its universal approximation ability, i.e. it does not cause instability
when Δt is relatively small [18].

1.2.3 Direct adaptive control in the presence of

bias
Consider adding a biased nonlinearity, with , to the
previous model
 (1.10)

In this case the leakage term multiplied by ν in (1.8) typically pushes

the weights in the wrong direction and performance suffers;
deadzone provides a more practical robust modification

(1.11)

where includes the maximum approximation error of the neural

network within its domain , combined with maximum external-
disturbance amplitude. Thus, the amount of system-plus-
disturbance knowledge required may make the method impractical
in many applications.
Many papers in the literature suggest adaptive-parameter
projection instead, even simple versions like

(1.12)

The designer requires knowledge of the maximum and minimum

weight for the i cell, and respectively, presumably
identified during pre-training.
 Without a priori knowledge of the system and disturbances,
learning control becomes the logical choice in the context of biased
nonlinearities. A direct adaptive control using supervised leakage,

(1.13)

can incorporate a learning control scheme when the weights that

estimate the bias are output from an online algorithm attempting
to identify the bias in real time; for stability reasons should only
be updated once. The robust term contains multiplying parameter
denoted because in subsequent sections will show this term
has similar effects as a control-bias term (a setpoint or feedforward
term).

1.2.4 A graphical model of personalities

The graphical model of personalities consists of a quadrant with
empowerment/manipulation on the horizontal axis, indicating how
one tries to affect the world and people, and emotion/logic on the
vertical axis indicating how one makes decisions. Manipulation
includes commands, requests, charm, and all subtle forms of the art.
Empowerment includes running away, avoiding, giving away power
—and at higher levels of thinking includes creating safe structures,
teaching, learning, planning, among many other things that do not
clearly come across as manipulation. The vertical axis denotes the
emotion and logic. Emotion provides coded memories, felt rather
than remembered, for quick access and for giving useful direction.
By logic we mean deductive logic, based on a set of beliefs; recall logic
qualifies as either valid or invalid, whereas beliefs classify as either
consistent or inconsistent. For instance, two people can disagree all
while following perfectly logical arguments based on their own
beliefs. Also, people only have a (perhaps delusional) self-image of
performing valid logical deductions and predictions.
Eight of the clinical personality disorders (PDs) anchor the system,
since they exhibit widely-agreed-upon qualities. Histrionic and
Borderline lie in the upper-right (UR), Avoidant and Dependent in
the upper-left (UL), Schizoid and Schizotypal sit in the lower-left
(LL), and finally the lower-right (LR) has Narcissistic and Antisocial.
The characteristic on the axis to the counter-clockwise for each
personality disorder seems quite apparent to all as the main
behavior; everyone experiences URs as emotional, ULs are seen to
empower others by leaving them alone or surrendering all their
power, LLs are seem overly logical when they are creating theories
(even if the underlying belief system seems unique), and LRs have
reputations for their shameless manipulation. The hidden
characteristic becomes apparent as a main motivator typically only
to people who engage in intimate relationships with PDs. Inside the
circle indicated in Fig. 1.6, better-balanced people have personality
self-images rather than PDs; for personalities the visible and hidden
characteristics predict biases rather than behaviors. The visible axis
indicates a projected bias, felt as an irrational hope; we describe the
hidden as a reflected bias, felt as an irrational fear of others exhibiting
this characteristic. Note that one does not necessarily move outwards
on the graph to a PD in one's quadrant if one was having mental
health issues [19], [20], and our model does not yet predict
movement on the graph. Personality theories that indicate self-
images include Lowry Colors and the Enneagram.
The Lowry True Colors system [21] provides personality
descriptions that closely match the mentioned qualities (Fig. 1.6).
Briefly, the Green personality types feel primarily logical; many
engineering undergraduate students feel Green. The Gold
personality types feel mission-oriented and like to empower others
through sharing knowledge and meticulous planning; many
teachers experience themselves as Gold. The Orange personality
types feel action-oriented and like to get things done by leading
(manipulating) others; many managers perceive themselves as
Orange. The Blue personality types feel primarily people-oriented
 g p y yp p y p p
and focus on building relationships using their emotions; many
empathy-based therapists experience themselves as blue.
The Enneagram has found vast popularity with the general public
and has nine types. If we move the types 1, 4, and 8 into new
positions it achieves consistency with the quadrant system. Full
qualitative descriptions of how the personality disorders, Lowry
Colors, and Enneagram personalities fit on the self-image quadrant
appear in [1]. Also by mapping Merrill's C.A.P.S. theory of work
place personalities [22] onto the quadrant, when considering it as the
way others perceive someone. For example, opposite types in the
Enneagram come across similarly to the outside world because
opposite self-images also imply opposite unconscious and opposite
unconscious communications.
The quadrant matches closely to how people in Western society
anthropomorphize some animals; a biological explanation for
personalities apparently becomes a possibility if we imagine that,
earlier in evolution, empowerment may have started with just passive
behavior. Manipulation may have started with simple active (Fig. 1.5)
movements. What if we consider animals as a type of robot,
engaging with the world through feedback? Then active may
indicate closed-loop control, and passive open-loop control. For a
human example, think of an athlete learning a new movement; at
first it is an active movement that requires closed-loop control
relying on feedback, with resulting excessive effort. Constant
practice and training will eventually turn the movement into a
memorized passive movement using open-loop control, resulting in
little (minimal) effort due to relaxed muscles.

1.2.5 A computer model of personalities

When the sensors provide measurements of the system error e and
its derivative , most industrial applications use a traditional
PID+bias control
 (1.14)

where the constants , , and defines the control gains, and

defines the control bias. In order to design optimal gain parameters,
one must choose a cost-functional. The reader may already know
how to use an LQR cost-functional to pick optimal PD gains; here we
use a nonlinear quadratic regulator cost functional based on a
framework of personality self-images

(1.15)
where the words indicate overall norm measurements e.g.
norms; the first two terms equate to the terms multiplied by the Q
matrix in an LQR control (assuming an SISO system with position
and velocity states), while the third term penalizes the PID control
effort for the first half of the step response, but for the second half
acts like traditional LQR only penalizing a measure of PD feedback
effort, e.g. . The term stems from understanding the
integral term as trying to compensate for the bias (or inaccuracies in
bias compensation). Thus, an integral qualifies as a simple learning
term, and its error measures the distance from its ideal value (the
bias that would result in as the error reaches zero at steady
state). We point out the following similarities to the qualities in the
graphical theory of personalities:
which allows a high-level control design e.g. choosing a desired
personality results in the weighting parameters in a nonlinear
quadratic regulator (NQR). This eliminates the need to pursue trial-
and-error, rule-of-thumb, or experiential choices of Q's and R's,
allowing an intuitive understanding of human personalities to guide
the design process at the highest level (Fig. 1.7).
In previous work we pointed out similarities in PID+bias control
to the Myers–Briggs personality behavioral techniques for humans:

and this model (Fig. 1.3) allowed us to predict the probability

distribution of Myers–Briggs conflict types (J/P and T/F) in the
human population. The optimization problem only requires three
parameters, so the four-parameter PID+bias is overparameterized; in
terms of human personalities we view this as free will for an
individual. Thus, the optimization must start with one a priori
parameter which we will refer to as our free-will parameter.
 FIGURE 1.3 Feedback model used to explain the
probability distribution of J/P and T/F in the human
population in [2].

The designer can first choose a personality, a precise angle on the

personality quadrant, and then choose a personality imbalance for a
magnitude. Consider using a parameter p to define the ratio of
personality characteristics on opposite sides of the graph: for
example in Fig. 1.8 the designer has chosen a Type 5 personality and
by choosing and the other qualities stem from
the chosen imbalance, and with .
Consider the proposed precise design for the cost-functional

(1.16)

where and defines the learning error of

the integral term. The integral effort gets penalized during the first
half of the step response, but after that we worry only about its
learning error. PD control defines the pure feedback component
while the memory (integral) term uses
initial condition . The algorithm sets an initial condition for
the integral term as in the first run and then at for the
second. Note the first term simply uses traditional LQR design; the
NQR simply adds a component that encourages the integral's
learning error to draw near zero by time .
Thus, a nonlinear quadratic regulator producing an optimal
PID+bias control models how humans choose a behavioral technique
(Myers–Briggs) to try and achieve consistency with their self-image
personality (Enneagram); since the model stems from the principles
of feedback, the assumption that evolution did not find a way to
escape the principles of feedback implies this model may indeed
capture the underlying principles of human personalities. Modeling
the other Myers–Briggs letters with feedback (Fig. 1.4) allowed us to
predict the probability distribution of N/S (given I/E) in the human
population, and the interested reader might refer to [23] for more
detailed explanations.

FIGURE 1.4 Control-system model from [3] used to predict

I/E and N/S distributions.
FIGURE 1.5 Personality self-images of Lowry True Colors
can be understood by anthropomorphizing animals. Green
ellipse: many engineering students may identify with Orcas
who learn logical hunting techniques, which may be called
passive since they are very careful about what they eat.

FIGURE 1.6 The clinical personality disorders represent

unbalanced people. Green ellipse: a shizoid tells people
about their theories (visible logic), with the idea of warning,
helping, or enlightening others (hidden empowerment).
 FIGURE 1.7 The proposed new Enneagram configuration
moves types 1, 4, and 8 to make it consistent with the
quadrant system. Noticing similarities between the four
personality qualities and Q1, Q2, R1, and R2 in an NQR
control results in an intuitive control-system design
process. The oval gives way to visualize the self-image of
a Type 5, for example, who has all four qualities but not in
equal measure.

1.3 Proposed methods

1.3.1 Proposed learning law
Based on our previous observations we propose using supervised
leakage

(1.17)

where the bias weights start at zero, , and

 (1.18)

where the moment in time the algorithm captures a bias estimate,

i.e. when the system first begins to remain close to the origin with
close to zero velocity or, equivalently, a settling time. We only expect
to capture an estimate of the bias, and the continued updates make
up for inaccuracies in this bias estimation and compensate for more
detailed characteristics of the nonlinearities near the origin.

1.3.2 Cost functional for optimization

Using the similarity between the four quadrant axes and the
standard LQR problem, we propose the cost functional

(1.19)

The first three terms look familiar, known from the LQR cost
functional, while the feedback effort only includes

(1.20)

and not the neural network outputs (Fig. 1.9).

 The total extra effort should penalize the effort needed above a
positive bias, but not any effort below the bias. Here we look at this
term as an opportunity to also limit weight drift; rather than
measure neural network output directly we look at the level of
individual weights, and penalize each weight for being too large-
positive. Otherwise, a weight activated only in a region where the
basis function had a small value wouldn't get penalized enough for
its growth; the proposed method ensures each weight gets equally
penalized for growing regardless of the value of the basis function.
This measurement of extra effort gets updated at time step k, of
length Δt, according to

The learning error denotes the difference between neural network

portion of the control and all the nonlinear terms; describing by
functions of time gives the expression

Unlike an LQR control, the design with an integral must consider

initial conditions. Thus, we run the step response twice; the first has
all weights initialized to zero and the second keeps the same weight
values from the end of the first run. Thus, the total cost functional
becomes
 (1.21)
with initialcondition for the two step responses
, where and ; the simulation in the
Results section uses .

1.3.3 Stability analysis

This section investigates stability theory in terms of Lyapunov
functions. The optimization values , , and will dive form
the personality angle and magnitude; themselves functions of the
personality imbalance parameters (see Fig. 1.8 for an example). The
imbalance in the personality gets described by parameter p. Using
the Matlab® optimization tool fmincon() generates the optimal
values for , and with the proportional gain as input, i.e. a free
will parameter.

FIGURE 1.8 Example: choice of Type 5 personality with

imbalance parameter p = 0.5 results in particular values for
the NQR cost-functional weightings.
 FIGURE 1.9 Computational model stemming from Fig. 1.3.
NQR models Enneagram and PID+bias models MB
conflict types (J/P and T/F).

A Lyapunov candidate Eq. (1.22) enables stability boundary

identification when the designed control law (1.7) and the proposed
update term (1.13) gets applied to (1.5). The adaptive Lyapunov
functions

(1.22)

where indicates the weight errors, with column vector w

indicating the (unknown) ideal values of the weights, results in time
derivative

(1.23)

Those familiar with (supervised) leakage might recall that (1.23)

implies signals are uniformly ultimately bounded (see the
appendix).
Other documents randomly have
different content
“cracked a smile” at the queer sights. Tragedy surrounded them,
which would not allow of an expression of humor. There was nothing
funny about a town burning down.
Just as Rob had anticipated, they found the other fellows in the
vicinity of the deserted building that they meant to destroy, in hopes
of staying the onward sweep of the flames. Andy and Tubby,
remembering what their chums meant to do, had taken up their post
close by, and anxiously counted the passing minutes, even while
also observing the many strange sights that continually cropped up
around them.
Tubby had read about Bedlam, and he wondered if it could “hold a
candle” to the mingled noises surrounding them. He had listened to
the crackling of flames before now, for Tubby had seen some pretty
big fires both in the city and in the woods; but those in the past were
not to be compared with what he now experienced. While Tubby was
standing there, now resting on one foot, and again on the other, he
heard Andy give a whoop of delight, and, on turning, beheld the
welcome face of Rob Blake, as well as that of Sim, and back of them
the tall figure of Mr. Jeffords, all of them carrying something in their
arms.
 Chapter XIX
Behind the Broken Door
“You got what you went after, then, didn’t you, Rob?” asked Tubby,
delighted to see the scout leader again.
They had gone off at such a furious pace that ever since Tubby
had been anxious concerning their safety. While watching the
advance of the devouring element, and the desperate efforts of the
gallant firemen to try and throw some obstacles in the path of the
flames, Tubby’s thoughts would constantly go out toward the absent
chums, with whose well being he was so greatly concerned.
“Yes, it’s here along with us,” Rob told him.
“But, say, what about Ralph?” demanded Andy, suddenly noticing
that the party he mentioned was absent.
“He came back with us, all right,” the scout leader replied. “Just
now he’s trying to find the Chief of the Fire Department. You see, we
couldn’t start trying to blow up this building here without his consent.”
“Land’s sake alive! I hope they get here pretty quick, then!”
ejaculated Tubby. “It’s coming closer and closer right along. Andy
and myself were getting ready to clear out of here when you showed
up.”
Rob had already noticed this for himself. His nervousness came
back, only instead of a fear lest some accident happen to the
speeding car, he now had something else to bother him.
They could feel the intense heat where they stood. Sparks fell all
around them, endangering their eyes if they ventured to look upward.
“Better turn your coat collars up, fellows,” advised Rob. “If you got
one of these flying sparks down your back you wouldn’t like it much.”
“That’s right,” commented Andy, “an ounce of prevention is better
than a pound of cure. Whew! just look at that cloud of sparks rise up!
Must have been a building collapsed then to make the shower.”
 “But I don’t see Ralph coming along yet,” complained Tubby, as
he strained his eyes to stare down the street where the other must
appear if he came at all.
There were plenty of people in sight, most of them onlookers who
were fascinated by the dreadful picture presented by the roaring
abyss of flames, though others were hard pressed firemen dragging
their hose about in the endeavor to find some place where they
might throw a stream into that old building just in the path of the
oncoming flames.
Precious seconds were passing, Rob knew. He wondered whether
he dared attempt to carry out his bold plan even without the consent
of the Fire Chief. One thing certain, unless Mr. Megrue came along
shortly, the last chance to attempt saving the mills would be lost. The
fire was already close to the ruined building, and at any moment they
might expect to see it burst into flames, for it must by now be getting
frightfully hot.
Tubby gave a sudden cry, and was seen hurrying toward a wagon
that, loaded with household goods, and drawn by a man and woman,
both foreigners, in place of a horse, had just passed them by.
Something had fallen off the load; it sprawled there in the street, and
lusty yells announced that it must be a fat baby.
Tubby picked it up and hastened to overtake the man and woman
who were tugging so lustily between the shafts. Then, having given
over the child into the charge of its mother, Tubby gallantly took hold
himself, and started to assist the man to draw his possessions
further away from the grip of the fire demon.
That was just like Tubby, whose heart always beat in sympathy
with any one in trouble. He would do without a meal in order to enjoy
seeing a hungry dog devour what had been intended for his own
consumption.
He did not intend going any great distance. After tugging for half a
block at the load, he felt that he had done his duty. Besides, if Rob’s
plan worked as they hoped, this section of the town would be
spared, and there was no absolute necessity for any one fleeing to a
great distance.
 So Tubby came back again, blowing like a porpoise from his
exertions. Ralph had not shown up yet, but Rob placed a hand
affectionately on his shoulder and it needed nothing more to tell
Tubby that his act had both been seen and appreciated at its true
worth by the one whose good opinion he chiefly cared for.
“I believe they’re coming, Rob!” Andy announced, greatly to the
satisfaction of everybody.
It turned out that way, and almost immediately the others also
discovered the figure of their companion hurrying towards them. At
Ralph’s side strode the tall form of a man wearing a glazed fire hat,
and a coat that was supposed to be immune to sparks and flames.
“I guess that’s Chief Megrue,” remarked Mr. Jeffords, eagerly,
“though his face is so black with the smoke and dirt his own wife
wouldn’t recognize him. Yes, I know his walk. Now we may see
something done that will bring results. I was just about to suggest
that you go ahead on your own hook, Rob, and I would stand for the
consequences. Desperate cases require desperate remedies, you
know. But there’s no need of our doing that now, fortunately.”
Ralph piloted his companion to the spot where the others awaited
them. Undoubtedly he must have already found occasion to explain
Rob’s scheme to the head man of the local fire department. Mr.
Megrue had been instantly seized with a conviction that it was
indeed the only thing left to them, since all the puny efforts they had
made to try and drown the fire out had been without avail.
Every fireman must know that such things are resorted to in all big
city fires when the violence of the wind makes it impossible to head
the flames off in any other way. Sometimes it turns out successfully;
and then, again, there may be no tangible result, owing to
circumstances over which even the use of explosives will have no
control.
Rob saw with more or less pleasure that the Chief was a man of
action, and not one of those old fogies sometimes connected with
local companies, all “fuss and feathers,” but lacking in the dash and
valor that should distinguish every firefighter.
Perhaps had he known where he could put his hand on a store of
explosives, the Chief might before this have considered the
possibility of blocking the flames by destroying that old connecting
link of a building.
“He says it’s the finest thing going!” called out Ralph, as he drew
near where his father and the boys were waiting for him. “As I know
how to handle the stuff, as well as the battery, he asks us to do the
job. Rob, I want your help only. Come with me. The Chief will smash
in the door for us, and we can plant the charge in a hurry. Then it
only means running the wires out to some place here, and
afterwards you can touch her off. That ought to be your privilege,
Rob, because you thought up the scheme. Right along this way,
Rob, please!”
Perhaps Sim and Andy, not to mention Tubby, might have liked
very much to be given a part to play in the game; but their common
sense told them there was no necessity for more entering the
building than just those two, unless Chief Megrue chose to
accompany the boys. At least, they could have the satisfaction of
being present when the big affair came off, and watching that
ramshackle building collapse into kindling wood after the charge was
fired.
The Chief fortunately had a fire ax with him. It may have been
used to batter in the doors of many a building that had commenced
to smoke, and give other positive signs of being about to burst into
flames. Besides, it is always a useful tool in the hands of your
country-town fireman, with which he loves to smash windows, and to
make a grand show of accomplishing things.
Straight up to the door of the unused building the three of them
hurried, the boys bearing the battery, the little cask containing the
dynamite cartridges, and the coil of wire necessary to complete the
outfit.
“The door is locked or barred!” shouted Sim, which remark told
that he had, during the absence of Ralph and Rob, made a little
private investigation on his own account, for it happened that Sim
Jeffords was of a very inquiring nature.
This information saved the others from wasting any time trying to
open the barrier. It also gave Mr. Megrue his clue, for, rushing
directly up, he commenced to wield that ax of his in a way that
proved him to be a born fire-fighter.
 With every blow Tubby and Andy and Sim gave vent to a whoop;
which helped get rid of their extra enthusiasm, and also showed how
they appreciated the work of the axman.
“Go to it, Chief!” shrilled Sim, excitedly. “That time the chips flew,
I’m telling you! Give it another smash like that, and you’ll see it go
down. Wow! look at that crackerjack hit, will you, fellows? It’s an
extra stout door, all right, but just the same it’s got to go! Once more
to the breach everybody. So, there!”
The ax sank into the door, and ripped the panels up. Again did Mr.
Megrue swing it back of him, and once more was that smashing
sound heard, so pleasing to the ear of the listening and watching
boys, who kept one eye on the little drama being played before
them, and at the same time managed to observe the close approach
of the sea of fire menacing the deserted factory.
Finally a shout from Sim announced that the door had been
beaten open, and that the way into the building no longer held any
obstacle to the advance of the trio.
They vanished from view, Fire Chief, Rob and Ralph. The three
boys, together with Ralph’s father, had to possess their souls in
patience while waiting for the reappearance of the bold invaders. At
any moment now they fully expected the building to burst into
flames, so overheated must the dry woodwork have become. It was
an anxious time for Mr. Jeffords, who had seen his son disappear
beyond that broken door. At the same time he knew that Ralph was
capable of thinking for himself, and also that he and Rob would not
attempt any foolhardy feat just for the sake of “showing off.”
There was a method in their madness in the present case. The
welfare of the whole community hung in the balance, for if the fire
could not be stayed by this master stroke of genius, then was
Wyoming doomed; and the morrow’s sun would rise upon the
smoking ruins of the entire town.
Seconds passed, and changed into minutes. When it seemed as
though something like two of these had drifted along, every one was
growing weak with apprehension. They had terrible visions of some
evil having overtaken those who had gone into the building. It was
old, and the floors almost in ruins, so that there was even a
possibility that the venturesome ones had fallen through some
unnoticed gap. Then, again, how were they to tell if some sort of
insidious gas had gripped their chums, and rendered them helpless?
These were the kind of strange fancies that took possession of
Sim Jeffords, always gifted with a lively imagination. He pictured Rob
and his cousin lying there inside the old wreck of a factory, victims of
a treacherous gas, and soon to be devoured by the oncoming
flames.
Sim could stand it no longer. He must ascertain the truth, for it
was always better to know the worst, according to his policy. He
would rush forward and cautiously enter the building, groping his
way through the smoke-filled interior, and calling out the names of
his cousin and Rob.
Filled with this resolution, Sim had even commenced to approach
the open door, paying no attention to the calls of his uncle from the
rear, when, to his great relief, he saw figures bursting out of the
structure. They were Chief Megrue, Rob, the scout leader, and Ralph
Jeffords.
 Chapter XX
The Struggle to Save the Town
Sim counted the figures that came trooping out through the
opening where the smashed door lay a wreck. When he found that
none were missing, a great load seemed lifted from his heart.
“Hurrah! it’s all working out as it should, and now we’ll see what
Rob Blake’s plan is going to do for Wyoming!” he found himself
shouting; although Sim was so greatly excited, he hardly knew
himself what he said.
Ralph could be seen with the reel of wire in his possession. He
was plainly unwinding it as he went, while Rob undertook to see that
there were no kinks in the strands as they came off the barrel of the
reel. Soon they were alongside Sim, who fell in step with the others.
The Fire Chief looked additionally anxious. He undoubtedly knew
what tremendous issues were involved in this last effort to save the
town. Gallantly had he and his followers battled with the destroying
foe, but so long as that furious breeze held out, all their efforts would
have been in vain. He knew that unless some such desperate
remedy as this explosion came to the relief of the outclassed fire
department, the fate of Wyoming was sealed.
He devoted himself just now to seeing that there were no
stragglers close enough to the doomed building to be injured when
the dynamite was exploded. Several men belonging to one of the fire
companies could be seen nearby, trying to turn the nozzle of the
hose they wielded so that a stream of water would fall upon the old
building; for it was recognized that unless the conflagration could be
halted at this particular spot all work was useless.
Chief Megrue shouted to them, and when he had by this means
caught their attention, he beckoned wildly. Although they might not
quite understand what his motions meant, at least they were
accustomed to obeying orders without questioning.
 Dragging their hose after them with an effort, they commenced to
head toward the spot where the Chief stood. This promised to take
time, for their load was heavy; and seconds just then counted
considerably more than a new supply of hose meant to the imperiled
town.
“Drop it, and come here as fast as you can run!” bellowed Chief
Megrue, using his hands in lieu of a megaphone; for it seemed that
he had lost his trumpet while hard pushed to escape from a fiery trap
when caught in a building that suddenly burst into flames, giving him
scant time to dash out again to safety.
The men looked as though they could not understand what he
meant. They evidently disliked abandoning a length of perfectly
sound hose when there seemed no absolute necessity for it. Again
did the Chief shout at them, continuing his wild beckoning at the
same time.
“Hurry this way! Drop that hose, and come as fast as you can! We
are going to dynamite that building, don’t you understand, you
fools?”
Perhaps that one word “dynamite” gave them their cue, for without
any further hesitation the three men dropped the hose, which
commenced to writhe like a snake as the force of the water passing
through the brass nozzle turned it this way and that. So they started
to run toward the Chief.
Meanwhile Ralph, assisted by Rob, had continued to move
steadily away. The wire would allow of their proceeding to a safe
distance, when the battery could be quickly attached, and a
connection made.
Sim had now been joined by both Tubby and Andy. All of them felt
their hearts beating tumultuously, for no one could do more than
guess at the result of the bold plan.
One thing sure, they had not left the old building a minute too
soon. Why, it was on fire even then, for Sim could see flames
creeping along the roof. He wondered whether there would be any
break in the wire sufficient to prevent the full current of electricity
from passing along it when Ralph came to making the contact. If
such a thing did happen, it would be much too late to try and remedy
the defect, because after that time it must be sheer madness for any
one to dream of entering the deserted factory again.
Surely Ralph and Rob must by now have gained a sufficient
distance to be safe from any debris that might be thrown up into the
air when the explosion came. Still, Ralph was the doctor in this
instance, for he had had the experience in blasting out stumps and
dislodging great rocks by the use of the dynamite.
Mr. Jeffords was there, too, watching his boy with eagerness.
Then, besides, there were the trio of bedraggled firemen who had
joined their chief at his command, all of them deeply interested in the
success or failure of the scheme that meant so much for Wyoming.
The clamor all around them was just as boisterous as ever, for
certainly nothing had occurred to tone it down. The citizens, facing
complete destruction for their town, were hoarse with shouting; but
their main thought now was to make sure of the safety of the women
and children, who could be seen fleeing through most of the streets,
awed by the terrible nature of the calamity.
Nevertheless, none of those who watched Ralph and Rob paid the
least attention to these outside occurrences. The fire might roar, and
the populace shriek in terror, but just then their undivided attention
was given to the little drama being unfolded before their eyes.
Ah! Ralph no longer pushed on. Whether he had reached the end
of his wire coil, or believed that it was useless to cover more ground
did not matter. What did count was the fact that he and Rob were
feverishly attaching the battery and switch to the wires, showing that
they intended to test their hastily arranged programme without
further delay.
“Oh! hurry, hurry, please!” groaned Tubby, speaking to himself, of
course, but in so doing voicing the feeling that held sway in every
heart; for the building was rapidly being clutched in the ever
extending fingers of the fire, and would soon be such a mass of
flames that to wreck it might do more harm than good.
“It’s all ready!” whooped Sim, as he saw by the actions of the two
boys that they had succeeded in making the attachments.
Then Ralph pushed Rob forward. It could be seen that he meant
to force the scout leader to press the button that would complete the
work. Even in that dreadful moment Ralph did not forget that it had
been Rob’s brilliant idea, after all, and to his chivalrous mind it
seemed only right and proper that Rob’s hand should be the one by
means of which Wyoming might be saved.
Rob did not hesitate. This was no time for expostulation. He
proceeded to do as the other urged him. Those who stood by held
their breath with awe, watching the old deserted factory the while, as
though everything depended on the result of Ralph’s efforts.
They saw the building give a distinct shiver. Then the very earth
under them actually rocked as if an earthquake had taken place.
Tubby was thrown down, and the rest had more or less difficulty in
keeping on their feet, such was the concussion of air as the
dynamite went off.
They also heard a heavy roar, and then, wonder of wonders, the
old building seemed to collapse, to fall down like a house made of
cards. There was a horrible series of crashing sounds; some
remnants even fell close by the boys, though luckily none of them
received the slightest injury. Then, as the smoke shifted and the dust
clouds settled, they saw a great gap where the deserted factory had
stood.
The Fire Chief was awake to the possibility of the debris
presenting a new peril. He immediately gave orders to the three men
to make their way forward, and if the hose were still in condition for
use to turn the nozzle toward the wreck.
Then he hastened away to fetch up others, so that all efforts might
be concentrated on that one section. The fight for the rest of the
town must be made then and there. Already a heavy blow had been
struck to beat back the fire; and if only it could be held to the other
side, the valuable plants might yet be saved.
Soon men came rushing along. An engine turned a corner and
stopped before a hydrant, with men jumping this way and that in the
endeavor to get connections with the least possible waste of time.
The explosion must have terrified most people who heard it, since
they could not but believe it meant the destruction of some mill. It is
wonderful, however, how news is flashed along from mouth to mouth
during such scenes of excitement as this; and, doubtless, when
people learned that the shock came from a desperate effort to
confine the fire to the side of the town where it had started, their
hopes would take a new bound upward.
“Will it force its way across in spite of everything, do you think,
Rob?” asked Tubby, solicitously, as he stood beside the patrol
leader, and watched the working fire-fighters battling so manfully.
“I hope not,” he was told, in a reassuring tone. “You can see how
they’re trying to soak the wall of that nearest factory so it can hold
out against the heat when the test comes. Besides that they are
getting more water on right along. Here comes another company,
and from the way the people keep cheering them I reckon they must
belong in another town, and have been rushed here by special train.”
It afterwards turned out that Rob had actually hit the truth when he
made this guess. Word of the dreadful imperiling catastrophe that
had come upon Wyoming must have been flashed to neighboring
towns by telegraph, as well as an appeal for assistance. Such a call
is never allowed to pass unheeded in American communities, and
just as soon as they could get the right of way a special train with the
engine and firefighters aboard had been dispatched, with the order
to “burn the rails” in making speed.
There was need of every available man and machine. The wall of
fire had by now arrived at the gap, and gave positive signs of being
disposed to leap across in order to complete its work of destruction.
Men fought madly to restrain it. Those who held the various lines of
hose pushed forward until their faces were scorched by the heat, but
in spite of all this they persisted, and would not be denied.
“They’re holding it there, mark you!” shrilled Ralph in the ear of
Rob Blake, as all of them stood watching these exciting happenings,
their hearts almost in their throats, so to speak, such was the weight
of their anxiety.
Rob was encouraged. He began to believe that after all the
devoted firemen were going to come out victors in their fierce battle
with the element that had started in to ravage the whole town of
Wyoming.
He bent most of his attention on the buildings close by, for if one
of them suddenly burst into flames it would mean that the worst that
could happen was about to commence.
Vainly did the flames dart forth, fanned by that still raging wind,
and endeavor to reach the buildings that had been cut off from their
grasp by the destruction of the old factory. The more they tried, the
greater the efforts of those who held the water nozzles continued.
Wherever a certain point was threatened there would be a
concentration of splashing fluid, and the enemy retreated, baffled
again and again.
Rob noticed presently that each time the fire made this attempt to
cross the divide it seemed to grow just a little weaker. He knew that
the fuel by means of which it was maintained was being devoured at
such a rate that, given time, the efforts of the Department would
prove successful. Even now matters had reached such a stage he
believed the worst to be over; and that victory was in the air.
 Chapter XXI
After It Was All Over
“I guess it’s all over but the shouting!” exclaimed Sim, which
remark proved that he too must have noticed some of the same
signs detected by Rob.
“Oh! do you really believe that?” cried Tubby, a little look of relief
appearing on his face, which lately had been screwed up in all sorts
of lines denoting his strained feelings.
“Well,” explained Sim, with alacrity, “you can see for yourself that
so far none of the sparks have set fire to the roofs of the mills and
factories, thanks to the men who are guarding the same, armed with
buckets of water. If you watch carefully you’ll learn that the blaze
keeps on getting weaker right along. It’s burning itself out, I tell you,
Tubby. We win, and most of the glory goes to Rob here for thinking
up such a grand scheme.”
“Don’t you believe that!” exclaimed the scout leader, energetically.
“Nine-tenths of it ought to go to Ralph, you’d better say. Didn’t he get
the apparatus for blowing up that factory, and wasn’t it Ralph who
piloted the car back home and here again, going like mad? But
perhaps you had better not say anything more about that business,
boys; neither of us want to pose as heroes. After all, what we did
was only the most natural thing in the world.”
“Just what it was, and we’re fishing for no bouquets, either,
remember,” said Ralph, who was close enough to hear all this talk.
It was now close on midnight, but no one in all Wyoming would
dream of such a thing as going to bed. Some had been burned out of
house and home, while all the remainder had suffered such a shock
to their nerves that sleep was quite out of the question for them.
The danger was past, and every one could take a good long
breath of relief. Why, even the wind had commenced to die down
now, as though conscious of its defeat in trying to aid its companion
element in destroying the bustling town.
 There would be plenty of work for all the firemen until dawn,
pouring a continuous stream of water on the ruins of houses, and the
lumberyard as well, that had been destroyed. Even when additional
companies reached town, coming from more distant places, they
were set to work rendering assistance to the stricken community; for
those gallant fellows would have been sorely chagrined if compelled
to return home without a chance to even wet their hose for
neighborly accommodation.
The boys knew that presently they ought to pull out and head for
the farm; but as long as Mr. Jeffords did not insist on going, it was
not their business to say the word. They felt that they would willingly
remain there for an hour and more, with so much going on to see
and to hear.
It was Rob who began to exhibit signs of impatience, for, to tell the
truth, the scout leader noticed that the crowd around them had been
growing of late; and he also discovered that a number of the more
prominent men of the town seemed to be talking very earnestly with
Mr. Jeffords.
Now, Rob began to take the alarm. He fancied that these people
were talking about himself and comrades, because often he could
see them look their way, and once a stout gentleman whom Ralph
had told him was the mayor of Wyoming actually pointed in their
direction.
“Don’t you think we’d better be starting home, Ralph?” Rob went
on to say.
The other chuckled as though he could give a guess as to what
was bothering his friend. At the same time he shook his head in the
negative.
“We’ve just got to wait till dad gives the word, you know,” he
explained. “Now he seems to be pretty busily engaged. Just hold
your horses, Rob. The fire has been gotten fully under control, and
there’s nothing more to be feared from that source. Unless I miss my
guess, the business men of this town have a duty, and a pleasant
duty to perform. I don’t mean to spoil it all by running away, not on
your life.”
Rob looked uneasy, but he was powerless to do anything.
 “Oh! well, I suppose we’ll have to stand for it, then,” he muttered,
as if making up his mind to take some punishment.
Ralph actually stared hard at him. Really, he had never before run
across a boy just like Rob Blake. Most of the fellows he knew would
have been only too delighted to find themselves in the limelight, and
called a hero. Nevertheless, Ralph was determined that Rob should
not escape by running away.
“They’re coming this way,” announced Tubby, who with Sim and
Andy doubtless had already “sensed” what was in the air, for all of
them were smiling broadly, and casting significant looks toward the
plainly annoyed Rob.
Mr. Jeffords led the dozen citizens straight up to where the little
group of boys stood, while a crowd pressed close around to hear
what was said.
“This gentleman,” began Mr. Jeffords, “is Mr. Perkins, our worthy
mayor, and these others are business men of Wyoming, some of
them owners of the costly mills and factories that have been saved
from destruction through a brilliant and bold scheme. I have
explained to them that though the dynamite that saved the town
belonged to me, and while it was my son Ralph who drove the car to
the farm to secure it, still the conception of the grand idea was wholly
Rob Blake’s. They want to congratulate you, my boy, on your bright
thought, and to thank you publicly for having been so instrumental in
saving our town from utter destruction.”
Rob opened his mouth to protest, but the stout mayor waved to
him that he had a few words to say; so the boy shrugged his
shoulders and held his breath. If they were so insistent, how was he
to help himself? Nevertheless, he could not think it fair that more of
the credit should be placed at his door than was given to resourceful
Ralph, who, once given the hint, had engineered the whole thing.
“I want to thank you, Rob Blake,” said the mayor, squeezing the
boy’s hand in his chubby palm, “in the name of all our people. Only
for your quick wit and ready brain we might at this minute be
witnessing the destruction of all our thriving industries, and Wyoming
would have been as thoroughly blotted out as some of those poor
Belgian and French communities have been before the vandal guns
of the warring Nations. And while I am about it I mean also to thank
Ralph Jeffords because of the valuable aid he gave once you
conceived the scheme of blowing up that building. Yes, let me also
shake hands with these other three lads here who wear the khaki, for
I feel sure that had you not been present, one of them would have
discovered some way to accomplish the same purpose you had in
view.”
Even Tubby blushed at the compliment. Plainly, then, these good
people of Wyoming were of an entirely different mind concerning
those who wore the khaki. Not so very long since, according to
Ralph, and they had settled in their minds that scouts were an
undesirable organization in a community, just because the first
attempt to raise a troop had been a failure, owing to the wrong sort
of leadership, and the poor class of lads enlisted under the colors.
“It may be that you boys have heard about the failure that
overtook a troop of scouts started in our town some time back,”
continued the mayor, and at that Rob began to show much more
interest than when he himself had been the object of attention. “But
some of us are now of the opinion that we were wrong in
condemning all wearers of the khaki because our own lads failed to
equal our ambitions. I am sure I voice the sentiments of this entire
community when I say that after having this lesson brought home to
us we realize that we have made a great mistake. For one, I am in
favor of urging that a troop of scouts be started here in Wyoming
without delay, founded on the broad and noble principles of your
order; and here and now I pledge you that my two boys shall
become charter members. And I also hope very much that Ralph
Jeffords will consent to take this enterprise in hand. All in favor of
this say aye!”
There came a unanimous shout from the assembled crowd, who
had been listening in absorbed attention while the mayor was talking.
After that other gentlemen came up, desirous of shaking hands
with the boys who chanced to be visiting at the Jeffords’ farm.
Evidently the story of how the scouts had been instrumental in
tracking the lost child, and restoring little Caleb to his nearly
distracted mother may have been brought to the town, for several of
the men mentioned something about it as they warmly greeted Rob
and Sim and Andy and Tubby.
 With such a wonderful background as that still burning section of
the town the picture was a striking one, which Tubby felt sure he
would never, never forget. He was particularly pleased that they had
cornered Rob, because on more than one former occasion Tubby
had known the patrol leader to slip away after having been
instrumental in accomplishing things worth while, just because he
could not bear to hear people making him out to be more than an
ordinary boy.
On his part, Rob was delighted to learn of the change that had
come over the good people of Wyoming with regard to their opinion
of scouts in general. They had seen a light, evidently, and would
know from this time on that those who wear the honored khaki, if
they are true scouts, are bound to be a credit to the community in
which they dwell.
Ralph, too, looked pleased. To tell the truth, ever since the coming
of his cousin Sim, and his acquaintance with such a fine fellow as
Rob Blake, Ralph had been wondering whether it would not be worth
while to make another attempt to organize a troop in Wyoming, if
sufficient encouragement could be received from leading citizens,
and the men at Scout Headquarters were also willing.
Now it seemed that matters had been taken out of his hands by
this strange happening, and the peril that had threatened the town.
Instead of it being Ralph who spent his time in going around and
beseeching citizens to allow their boys to join in with him so as to
form a new organization, it was the mayor himself who gave the
word, and even promised that his two sons should enroll their names
among the very first candidates.
Plainly, then, Rob and his chums had been able to show what
scouts are worth to any community and had made a deep and
lasting impression on these people; it would not be long before their
visit must begin to bear fruit to the honor of the whole organization.
Now that this ceremony had been carried through, perhaps Mr.
Jeffords would be willing that they start for home. Rob was beginning
to feel that a bed would not be unpleasant, for all of them had been
up very early on the preceding day, and so much had taken place
since then that they were tired.
 Tubby, too, was commencing to yawn, which was a pretty good
sign that he could not hold out much longer. All of them, in fact, were
glad when Ralph’s father came back and announced that they had
better start out to find the car, which it was to be hoped had not been
taken by any of the fleeing people.
This worried Tubby somewhat, for he wondered whether they
would be able to hire a rig, or another car in case their own had been
taken. All anxiety on this score, however, was soon set at rest, for as
they drew near the spot where Ralph had abandoned the auto they
discovered it still there near the curb.
They quickly found seats, Sim occupying one of the extra ones.
So they started forth, and many times did Andy, Tubby and Sim
crane their necks to look back toward Wyoming; but the angry glow
had faded from the heavens by now, and stars were beginning to
show themselves here and there through the clouds.
“Well, after all, we’ve gone and forgotten Peleg!” announced
Tubby, presently. “Too bad, and I hope he doesn’t have to tramp all
of those ten miles out to the farm.” But Rob somehow found himself
believing that Peleg had purposely kept away from them.
 Chapter XXII
Laying Plans
“Did anybody happen to see Peleg after the fire broke out?” called
Ralph, over his shoulder, as he continued to pilot the big car, the
headlights showing him all inequalities in the road, so that he could
avoid most of the “bumps.”
“I did,” spoke up Andy, immediately. “Let’s see, I think it was just
about the time that fat mayor was going around shaking hands with
us, and giving us that taffy about his change of heart regarding the
scouts.”
“Then Peleg should have known we meant to clear out pretty
soon,” interrupted Tubby, slowly, “so if he had a particle of sense,
and really wanted to come back home in the car with the crowd, why,
seems to me he’d have hung around.”
“Well, he didn’t,” added Andy. “I saw him grinning as though
tickled half to death about something. Perhaps now it pleased him to
see that mayor grabbing our hands so,—well,—I might say
effusively. How about that, Ralph; would Peleg care if he saw you
being patted on the back, and made a hero of?”
“He might, and then again perhaps it was something else that
made him seem so happy,” replied Ralph.
The other boys may not have understood the real meaning of
those words, but Rob did. He knew Ralph was hinting to him that the
farm boy may have held back from joining them because he began
to feel ashamed of what he had done, and could not bear to face the
owner of the stolen stamps so soon after selling the packets to the
curio dealer.
Even that failed wholly to convince Rob. When he believed in any
one it was hard to make him change his opinion. Why should Peleg
seem so well satisfied with himself? Surely, the getting of a few
dollars, more or less, in a shady transaction too, of which he must
later on feel ashamed, would hardly cause him to appear so happy.
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebookmass.com