Page 221 - Autonomous Mobile Robots
P. 221

206                                    Autonomous Mobile Robots

                                for all i = 1, ... , n − 1. Then for every  > 0 there exists a δ> 0 (depending
                                on  ) satisfying δ   c 0 ≤|x 1 (0)|=|ξ 1 (0)|≤ c 1 , such that the trajectories of
                                                             0
                                the system in closed loop with the C control law
                                                   
     α 1    α n−1
                                                         x      x
                                                    u x 1 ,  , ... ,    if |x 1 | >δ
                                                         21     2,n−1
                                               u =       x β 1  x β n−1                   (5.20)
                                                          1
                                                                 1
                                                    0                   elsewhere
                                                  
                                                           n
                                converge to the set     ={x ∈ R | x ≤  } in some finite time T ∗ and remain
                                therein for all t ≥ T ∗ .
                                   At this point the reader may argue whether it is possible or not to let δ go
                                to zero, that is, what we can conclude about the (discontinuous) control law
                                                    
            α n−1
                                                          x α 1  x
                                                    u x 1 ,  , ... ,    if x 1  = 0
                                                          21     2,n−1
                                               u =        x β 1  x β n−1                  (5.21)
                                                           1
                                                                  1
                                                   
                                                    0 p×1               if x 1 = 0
                                Observe that the control law (5.21) is discontinuous at x 1 = 0 as a function of
                                x, but it is continuous as a function of t, since x 1 (t) = 0 only asymptotically
                                (if x 1 (0)  = 0, which is without lack of generality). Moreover, by hypothesis,
                                                    β i
                                                 α i
                                the variables ξ 2i = x /x tend to zero when t goes to infinity. Thus
                                                 2i  1
                                                              α n−1
                                                     α 1
                                                    x (t)    x 2,n−1 (t)
                                                     21
                                         lim u x 1 (t),  , ... ,       = u(0, 0, ... ,0) = 0
                                                     β 1       β n−1
                                                    x (t)     x   (t)
                                        t→∞
                                                     1         1
                                   As a consequence, the control law (5.21) is well defined and bounded,
                                along the trajectories of the closed loop system, for all t ≥ 0 and, viewed
                                                                                         0
                                as a function of time, is even continuous (i.e., it is at least C )as t
                                goes to infinity. Finally, using Theorem 5.6, with δ = 0, and assuming
                                that the conditions (5.19) hold, we conclude that the control law (5.21)
                                almost exponentially stabilizes the system (5.18) on the open and dense set
                                          n
                                  ={x ∈ R |x 1  = 0}.
                                Remark 5.6   The assumption x 1 (0)  = 0 is without lack of generality,
                                as it is always possible to apply preventively an open loop control, for
                                example, a constant control, driving the system away from the hyperplane
                                x 1 = 0 [32,33,47].
                                Remark 5.7   By a general property of one dimensional dynamical systems,
                                we conclude that the state variable x 1 = ξ 1 evolving from a nonzero initial
                                condition approaches the equilibrium x 1 = 0 without ever crossing it, that is,




                                 © 2006 by Taylor & Francis Group, LLC



                                 FRANKL: “dk6033_c005” — 2006/3/31 — 16:42 — page 206 — #20
   216   217   218   219   220   221   222   223   224   225   226