Analysis and Control of Underactuated Mechanical S
Analysis and Control of Underactuated Mechanical S
Analysis and Control of Underactuated Mechanical S
net/publication/260799309
CITATIONS READS
0 36
1 author:
Charles Simpkins
RDPRobotics, LLC
22 PUBLICATIONS 156 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Charles Simpkins on 28 April 2015.
Analysis and Control of tuated (essentially a system possessing Before the classifications can be
Underactuated Mechanical more degrees of freedom to be con- presented in detail, UMSs must be
Systems trolled than there are control inputs), described carefully. The third chapter
A. Choukchou-Braham, B. Cherki, and what the potential sources of presents UMSs from the Lagrangian
M. Djemai, and K. Busawon, Springer, underactuation can be. formalism. The authors state that two
New York, 2014, 138 pages. Typically UMSs not only render major advantages of the Euler–
A
nalysis and Control of Underac- many well-known results and proper- Lagrange formulation are 1) the result-
tuated Mechanical Systems is a ties for analysis and control invalid but ing equations are invariant and 2) they
timely book. Currently, there also exhibit undesirable properties such allow one to compute the evolution of
are several major research as nonminimum phase behavior. The mechanical systems as a result of
projects in robotics aimed at creating authors do a good job of pointing out applied forces. The equations are first
better solutions to the problem of loco- known characteristics that are helpful, derived for fully actuated mechanical
motion. One of the major challenges is such as the fact that UMSs do not satisfy systems, then for UMSs, and finally a
the underactuated nature of locomotion a known necessary condition for the comparison with nonholonomic sys-
systems. Another challenge is the broad existence of smooth time-invariant tems is drawn. The chapter then
and often seemingly unsystematic feedback stabilization. describes the various forms of partial
nature of not only the problem but also After discussing how challenging linearization that are applicable to
the literature. The authors attempt to and interesting these types of systems UMSs and finishes with common cases
deal with these issues. can be, the authors present an overview (the inverted cart-pendulum, acrobot,
The book consists of five chapters of several methods of control, such as and several others).
and four appendixes. Each chapter ends energy based (swing the system up into The fourth chapter details the two
with many excellent references. The first a linear region then switch to a linear main methods for classifying the
chapter is an introduction and historical controller), passivity based (swing into a UMSs and explores the possibility of a
literature review. There have been few desirable orbit then switch), output- link between them. The key steps of
attempts like this to assemble the litera- feedback, feedforward, small gains syn- each method are laid out, and several
ture and present a systematic framework thesis, and sliding mode control. examples are provided, further aiding
for the control of underactuated The authors state that since UMSs in cementing the reader’s understand-
mechanical systems (UMSs). Although are so varied in type, and are typically ing. For example, the CFD method
this chapter does not cover all of the vast addressed on a case-by-case basis, classifies every UMS into a combina-
number of techniques that are available, this book attempts to present a means tion of a tree structure, chain structure,
it does provide many references for by which to look for common pat- or isolated vertex. The authors provide
those not described. The focus is mainly terns and determine what appropriate a balance of mathematical develop-
on unified frameworks that include control method should be applied ment with clear explanation so that the
nonlinear deterministic approaches. rather than outlining every possible reader is not overloaded. Ultimately,
The second chapter covers generali- controller. The classification of the the two classification systems are
ties and the state of the art on the con- problem is a key step, and this chap- determined to be independent but par-
trol of UMSs. It discusses in depth the ter introduces the two main classifi- allel tools.
concept of what it means to be underac- cations: the control flow diagram
(CFD) methods of Seto and John
Digital Object Identifier 10.1109/MRA.2014.2299502
Baillieul and the analytical methods
Date of publication: 10 March 2014 of Reza Olfati-Saber. (continued on p. 87)
In the fifth and final chapter, the If the reader does not posess a back- collate it into a coherent formal meth-
authors describe control approaches ground in nonlinear stability and con- odology without limiting the final
to address all of the classifications trol when he or she opens this book, the approach to a single control strategy. If
within the CFD method. Historically, appendices are a good starting point for roboticists are going to solve the prob-
backstepping was the proposed design fully appreciating the contents of the lem of dynamic locomotion, it is likely
method originally applied to the tree book. They describe theory, analysis that they will have to employ the type of
structures only. The authors extend approaches, appropriate geometry, and formal but flexible methodology pre-
and develop control approaches for some warnings about destabilization. sented in this book.
the remaining structures in this chap- Analysis and Control of Underactu- —Reviewed by
ter. Both examples and results from ated Mechanical Systems is a rare work Alex Simpkins, Jr., Ph.D.
various applications are described, in that it attempts to take a literature full San Diego, California
once again clarifying the concepts. of disparate cases and techniques and