Geometric control of mechanical systems. Modeling, analysis, and design for simple mechanical control systems. (Q3155754)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 2129143
Language Label Description Also known as
default for all languages
No label defined
    English
    Geometric control of mechanical systems. Modeling, analysis, and design for simple mechanical control systems.
    scientific article; zbMATH DE number 2129143

      Statements

      0 references
      0 references
      18 January 2005
      0 references
      Riemannian metric
      0 references
      affine connection
      0 references
      Lagrange's equation
      0 references
      stability
      0 references
      averaging
      0 references
      Geometric control of mechanical systems. Modeling, analysis, and design for simple mechanical control systems. (English)
      0 references
      The aim of this text is to provide a differential geometric frame for handling mechanical control systems. The Riemannian metric arises from the kinetic energy of the system (it can be represented by a matrix with suitable entries like masses and inertias) and for the associated affine connection, the solution to Lagrange's equation is precisely a geodesic. The field stands at the intersection of three big disciplines, namely differential geometry, mechanics and control theory, and the book aims at introducing to all three disciplines. The first part is conceived as modelling, with chapters on linear algebra, differential geometry (the definition of a connection, so fundamental for this book, repeats the one from many textbooks, but it is not the clearest as confirmed by the need for the remarks two paragraphs later; an intuitively satisfactory definition can be found in the book [\textit{J. P. Aubin} (ed.) et al., Mathematical techniques of optimization, control and decision. Annals of the CEREMADE, 1. Boston-Basel-Stuttgart: Birkhäuser. VIII (1981; Zbl 0472.00007)]), and the notion of a ``simple'' mechanical (control) system. Such systems are composed of interconnected rigid bodies, and the configuration manifold is built from \(\text{SO}(3)\times \mathbb{R}^3\) corresponding to rotations and displacement of the individual bodies. The free case leading to the above-mentioned geodesics is refined when external forces are present and next, when velocity constraints are present. The velocity of an admissible trajectory belongs to a certain distribution \(D\). When \(D\) is integrable, the constraints are holonomic, else, they are nonholonomic. Control inputs can also be added. They enter the system like forces as additional terms in Euler-Lagrange equations. The last chapter in the first part focuses on how to express symmetries in a differential geometric context. This is motivated here by rigid body kinematics, and then Lie groups and Lie algebras are introduced, together with invariant metrics and connections, group actions, symmetries, conservation laws, isometries and reduction. The second part studies fundamental issues in control systems applied to the kind of mechanical systems the authors consider here. General results on stability are cast in the mechanical context: the Lyapunov function is obtained from the potential from which the force derives. Controllability results for simple mechanical systems of this book (with force deriviing from a potential and velocity constraints) are also obtained from previous (mechanically) unspecific approaches of Stefani, Sussmann, Basto-Goncalvez, and others. The results of the authors are expectedly partial, and not all the proofs are presented completely. The reader is sent to previous publications of the authors. So the mechanical context does not alleviate much the fundamental difficulties associated to controllability studies. Another chapter is devoted to averaging studies when the controls are periodic signals. The authors construct an averaged mechanical control system, and more specifically an averaged potential is obtained if the controls derive from a potential. Series expansions for the solutions are computed. Still averaging procedures, like optimal control, seem sometimes like diversions, once one failed in the hard core of control theory (which is extremely hard to handle). The third part is more oriented towards concrete results and design. Stabilization of mechanical control systems is investigated. (The authors write a scary sentence on p. 481: ``The problem of stabilization consists, roughly, of finding controls that will steer the system to a desired state''. No, not ``roughly'', not at all, and this is not the system which is steered). In the mechanical control case, the control will shape the potential leading to a stable closed-loop system. Control Lyapunov functions are used. Finally, some results on tracking using oscillatory controls are presented, including the underactuated case. Many references, a symbol index and a subject index are included at the end. The reviewer feels somewhat uncomfortable with the architecture of this voluminous book. The ``modelling part'', the mathematical background (here believing that mathematics consists in modelling, while it is a language which expresses realities and might well lead to an imposture), presented ``hastily'' to quote the authors, takes a considerable portion of the volume, and it helps supporting specialized research results of the authors. So the student will be invited through a tour to the basics of differential geometry so that he will be able to understand what the authors are doing in their research. Would not it be preferable to assume a background in differential geometry, present a self-contained essay on one's resarch, and avoid a ``differential geometric brain storming'' which is likely to happen if this book is used as a textbook for a university course? Still the formulation of the mechanics in the Lagrangian picture, with the affine connection is the key which articulates the mathematical background to the differential geometric control methods. In this book, mechanics is gaining from the inputs of differential geometry and control theory. But control is not enriched in return and even less differential geometry. If this had happened, the book would have had another dimension. Obviously, systems science has a lot to learn from mechanics, and the Humboldtian and Newtonian pictures are still kept separate. Optimal control is not addressed here nor symplectic geometry, and this on purpose without valid justification, but this is a direction for great achievements!
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references