achieved with a relatively large margin. However, the specified maximum values were

pressed to their limits during occasional trials.


 Position Controller Performance

 Position control for each of the joints was critical because changes in the position of a

joint during the picking operation could have been the difference between successful and

unsuccessful attempts. As described chapter 7, this factor was the foundation for tight position

requirements. The performance of the position controllers was quantified by examining the

response of each joint to a step change in the position setpoint. Since the largest joint control

signals were generated by large position errors, maximum position changes were used for

generating the position step changes.

 Operation of joint 0 was confined to positions between 29.3 and -32.7 deg by the

intelligence base. To quantify the performance of the position controller for joint 0, a step

change of 51.8 deg was chosen. In this case, the setpoint was changed from -25.9 deg to 25.9 deg

while recording the reaction of the joint. The response of the system to a step change in the

position setpoint while joint 2 was centered in its support is shown in Figure 9.9. The joint

moved up to a maximum position of 31.5 deg before returning to a steady-state position of 26.0

deg. Thus, an overshoot of 11 percent (5.7 deg) was recorded along with a steady-state error of

0.1 deg. The worst case encountered in the trials occurred when the joint was excited by a

negative step with the end-effector in its extended position (Figure 9.10). The joint responded

by dropping to a position of -32.4 deg (13 percent overshoot). A steady-state error of 0.5 deg was

evident when the joint was returned to a steady-state position of -26.3 deg. Even these worst

case measurements were within the allowable range of operation for the position controller of

joint 0. From all of the trials, average values of 0.1 deg steady-state error and 11 percent

overshoot were found.

 The position of joint 1 was limited in the software to 15.6 degrees from its centered

location. Thus, the step tests for quantifying the performance of the position controller were