Page 277 -
P. 277

260                                                 REFERENCES

                     s instead of 2 s. Plot as well the Cartesian position (x 2(t), y 2(t)) of the end
                     effector in base coordinates.
                  (b) Redo the simulation deleting the lines in Figure 4.5.7 that zero the initial
                     velocity estimates.
                  (c)  Try to simulate the digital CT controller using the alternative technique
                     to compute  k from   and v k, as given in equation (1) in the example.
            4.5–2  Digital Control Simulation. Convert the PD-gravity CT controller in Example
                  4.4.3 to a digital controller. Try several sample periods.
            4.5–3  Error Dynamics for Digital Control. Find the error system in Table 4.4.1
                  using digital control of the form (4.5.3).
            4.5–4  Antiwindup Protection. In Example 4.4.4 we saw the deleterious effects in
                  robot control of integrator windup due to actuator saturation. In Example
                  4.5.2 we showed how to implement antiwindup protection on a simple PI
                  controller. Implement antiwindup protection on the robot controller in
                  Example 4.4.4. The issue is determining the limits on the integrator outputs
                  given the motor torque limits. Successful and thorough completion of this
                  problem might lead to a nice conference paper.
            Section 4.6
            4.6–1  Optimal LQ Outer-Loop PD Gains. Verify (4.6.15). To do this, select the
                  Riccati solution matrix as





                  with P i    R  Substitute P, A, B, Q, R into the Riccati equation (4.6.6). You will
                          n
                  obtain three n×n equations that can be solved for P i . Now use (4.6.5).
            4.6–2  Robust Control Using LQ Outer-Loop Design. Redo the PDgravity simulation
                  in Example 4.4.3 using PD gains found from LQ design. Does the LQ
                  robustness property improve the responses found in Example 4.4.3 using
                  nonoptimal gains? Try various choices for Q p , Q v , and R, both diagonal and
                  nondiagonal.
            Section 4.7
            4.7–1  Direct Cartesian Computed-Torque Design. Begin with the Cartesian
                  dynamics in Section 3.5 and design a computed-torque controller. Compare
                  it to (4.7.10).
            4.7–2  Approximate Cartesian Computed-Torque. Derive the error system dynamics
                  associated with the approximate control law (4.7.11).
            4.7–3  Cartesian PD-Plus-Gravity Control. Repeat Example 4.4.3 using Cartesian
                  computed-torque control, where the trajectory is given in workspace
                  coordinates.








            Copyright © 2004 by Marcel Dekker, Inc.
   272   273   274   275   276   277   278   279   280   281   282