Page 216 - Autonomous Mobile Robots
P. 216

Stabilization of Nonholonomic Systems                      201

                                (iii) The matrix function g 22 (x 1 , x 2 ) depends on x 2 only, that is,
                                    g 22 (x 1 , x 2 ) =¯g 22 (x 2 ) for some function ¯g 22 (·).
                                (iv) There exists a smooth vector function u 1 (x 1 , x 2 ), zero for x 1 = 0
                                    and for all x 2 , that is, u 1 (0, x 2 ) = 0, such that

                                                 −∞ < x Xg 11 (x 1 , x 2 )u 1 (x 1 , x 2 )< 0
                                                        1
                                    for some positive definite matrix X and for all nonzero x 1 in U.
                                                                                        ¯
                                                                        ¯
                                    Moreover g 21 (x 1 , x 2 )u 1 (x 1 , x 2 ) is smooth in U and it is a function
                                    of x 2 only, that is, g 21 (x 1 , x 2 )u 1 (x 1 , x 2 ) = f 2 (x 2 ), for some function
                                                                      ¯
                                    f 2 (·) such that f 2 (0) = 0.
                                                ¯
                                     ¯
                                 (v) There exists a smooth function u 2 (x 2 ) that renders the equilibrium
                                    x 2 = 0 of the system
                                                    ˙ x 2 = f 2 (x 2 ) +¯g 22 (x 2 )u 2 (x 2 )
                                                        ¯
                                    locally asymptotically stable.

                              Then, the smooth control law


                                                                u 1 (x 1 , x 2 )
                                                 u = u(x 1 , x 2 ) =
                                                                 u 2 (x 2 )
                              locally asymptotically stabilizes the system (5.12).
                                 As should be clear from Theorem 5.3, the possibility of rendering the
                              manifold x 1 = 0 invariant for the closed loop system, allows the asymptotic
                              stabilization problem to be solved in two successive steps. Hypothesis (iv)
                              determines the component u 1 of the control law; whereas the component u 2
                              must be chosen to fulfill hypothesis (v). Observe that the choice of u 1 is crucial,
                              as the existence of a smooth function u 2 (x 2 ) fulfilling hypothesis (v) depends
                              on such a choice. The hypotheses of Theorem 5.3 may be easily strengthened
                              to obtain a global result.

                              Theorem 5.4  Consider the system (5.12) defined in an open and dense set
                                               n
                                         ¯
                              U, such that U = R . Suppose (i), (ii), (iii), and (iv) of Theorem 5.3 hold.
                              Moreover, suppose that the following holds:

                                 (v) There exists a smooth function u 2 (x 2 ) which renders the equilibrium
                                     x 2 = 0 of the system
                                                         ¯
                                                    ˙ x 2 = f 2 (x 2 ) +¯g 22 (x 2 )u 2 (x 2 )
                                     globally asymptotically stable.



                              © 2006 by Taylor & Francis Group, LLC



                                 FRANKL: “dk6033_c005” — 2006/3/31 — 16:42 — page 201 — #15
   211   212   213   214   215   216   217   218   219   220   221