|
In this thesis, we propose using three different kinds of
measures according to kinematic, dynamic, and energy senses to
evaluate manipulator postures with respect to compliance task
requirements. In previous researches, the measures were usually
used to determine optimal postures in advance, and the
improvement to system performances generally examined in a
static manner. We show that postrues derived in static sense may
fail in controlled dynamic systems; in the other word, the
merits (deficiencies) caused by good (bad) postures can be
compensated for by the controller. However, postures determined
according to proper performance measures can still alleviate the
burden of the controller and be helpful to the control strategy
realization. On the other hand, different postures may be used
in different phases of the compliance task. Thus, a posture
selection and control scheme is proposed to govern the sequence
of postures selected according to task requirements and
environments. We develop the proposed scheme based on the
redundant robot manipulators, because of their better
manipulability in using complex task execution. Simulations that
demonstrate the effectiveness of the proposed scheme are
described.
|