Page 172 -
P. 172

154                                                Robot Dynamics

              Independent Joint Dynamics. In many commercial robot arms the gear
            ratios r i  are very small, providing a large torque advantage in the actuator/
            link coupling. This has important ramifications that greatly simplify the design
            of robot arm controllers.
              To explore this, let us write the complete dynamics by components as


                                                                        (3.6.7)



            where B≡diag{B i } and d i  is a disturbance given by


                                                                       (3.6.8)


            with m ij  the off-diagonal elements of M’, V jki  the tensor components of
            the friction of the ith link, and G i  the ith gravity component.
              This equations reveals that if r i  is small, the arm dynamics are
            approximately given by n decoupled second-order equations with constant
            coefficients. The dynamical effects of joint coupling and gravity appear only
            as disturbance terms multiplied by  . That is, robot controls design is virtually
            the problem of simply controlling the actuator dynamics.
              Unfortunately, modern high-performance tasks make the Coriolis and
            centripetal terms large, so that  d i is not small. Moreover, modern high-
            performance arms have near-unity gear ratios (e.g., direct drive arms), so
            that the nonlinearities must be taken into account in any conscientious controls
            design.
            Third-Order Arm-Plus-Actuator Dynamics

            An alternative model of the complete robot arm is sometimes used in controls
            design [Tarn et al. 1991]. It is a third-order differential equation that should
            be used when the motor armature inductance is not negligible.
              When the armature inductances L i are not negligible, instead of (3.6.2) we
            must use the armature-controlled do motor equations

                                                                       (3.6.9)



                                                                      (3.6.10)

                    n
            with I   R the vector of armature currents,




            Copyright © 2004 by Marcel Dekker, Inc.
   167   168   169   170   171   172   173   174   175   176   177