Summary: | 碩士 === 國立成功大學 === 電機工程研究所 === 83 === Previously, joint-angle limits of a redundant manipulator have
been formulated into the inequality-constraints of the
quadratic programming(QP) formulation. And, the corresponding
joint-limit avoidance problem has been solved by applying the
Compact QP method. However, recently, we have discovered that
this approach will cause discontinuities in the joint
velocities and, further, result in unacceptable accelerations.
In order to resolved this problem, the task of joint-limit
avoidance will be switched from the inequality constraints into
the objective function. At the same time, the original
objectives, such as singularity avoidance and drift-free
consideration, shall also be preserved. Therefore, multiple-
goal considerations shall be conducted. In this research, a
systematic algorithm, namely the Parallel Scheme will be
proposed to solve the multiple-goal problem. It can adaptively
adjust all the weighting factors of the corresponding goals
according to their individual needs under dynamically varying
environments. The 7-DOF CESAR redundant manipulator will be
used as an example mechanism to test the proposed algorithms
for joint-limit avoidance and multiple-goal considerations.
|