Nonlinear control and analytical mechanics. A computational approach. With 1 CD-ROM (Windows, Macintosh and UNIX) (Q1841307)

This book is a rather general presentation of recent developments in nonlinear control with an emphasis on differential geometric methods and mechanical systems. The goal is to make the reader familiar with the concepts by combining analytical insights and symbolic computations. Chapter one is a standard introduction to ordinary differential equations (qualitative theory, stability analysis using the Lyapunov theory). Chapter two is an introduction to the differential geometric background used to study nonlinear control systems. This background can also be found in recent books of Isidori and Sastry. One starts with manifolds, vector fields, and one continues with distributions (with a statement of the Frobenius theorem) and one ends with Lie groups and Lie algebras. The authors look at their objects in coordinates. Let us observe that their definition of vector fields (by substacting nearby points in a manifold) is nonsense in general. Moreover, their introduction of the derivation concept later is not convincing (to put it mildly). The authors should have read more carefully the classical book of Warner which they cite afterwards. That the author's understanding of a vector field is shaky is later confirmed when they add two (invariant) vector fields on a Lie group to conclude that they form a vector space (p. 83) and later only do they mention that the invariant fields can be identified with the tangent space at identity. Chapter four deals with the kinematics of multibody mechanical systems. They focus their attention to the case when a tree structure is formed. In chapter five, the Hamiltonian dynamics of holonomic systems is obtained from the Lagrangian. Then the case of nonholonomic constraints is examined. Some simulations, examples and considerations of a mechanical nature (especially for chain and tree configurations) are added. Chapter six is on nonlinear affine control systems and one studies controllability, observability, Kalman type decompositions, linearization, inversion, regulation and dynamic extension. The tools are those of chapter two. Chapter seven deals with robust and adaptive control. The nominal systems are feedback linearizable and one studies under which conditions on the perturbations, one gets a transformed system in suitable normal form. The strongest is the matching condition which is investigated later in a Lyapunov type design. Next, one deals with robust stabilization via backstepping for weaker forms of the perturbation (triangular structure). Adaptive versions are considered at the end of the chapter. The last chapter is an overview on variable structure control with robustness considerations. The backstepping method is reexamined in this context. Each chapter has its list of references and problems. There is one appendix (Appendix A) which provides information on ``Propac'' a Mathematica type package for symbolic computations and a medium sized index. A CD-ROM disc is included and it allows to use ``Propac''. The chapters can largely be read independently and they explain known facts. This juxtaposition of topics diluted in examples and programs for the computer lacks internal harmony. This is not the kind of book which is intended to be read in one's bed. Instead it is expected to sit behind a computer and play with ``Propac'' increasing continuously one's practical feeling for the various concepts introduced. This leads to what could be called ``computer assisted reading''. This formula, which is the credo of the authors and the only originality of the volume, might be of value for advertising recent developments in differential geometric control so that practically oriented minds can realize that one can do things with this theory. However, the real impact in applications is still unclear. Let us finally quote the authors: ``Mathematicians may decide that our arguments lack rigor and some engineers may find our discussion too formal. However, we can judiciously sacrifice rigor for accessibility, but we often need precise statements to clearly identify the range of applicability of a technique or to establish reliable machinery for computing.'' So the authors find just as well a convenient niche where they avoid the hardships of rigor without having to get their hands too dirty with real implementations. Once more, it remains to see if this way of dividing applied science at its own advantage is always meaningful.