Robot control strategies utilize the dynamic robot model to decouple the joints and specify the closed-loop response. The implementation of model-based control algorithms, however, is based upon the on-line evaluation of robot dynamics. This evaluation, which is on the order of milliseconds for typical industrial manipulators, introduces discretization errors (in addition to potential modeling and parameter errors) and degrades performance. In this paper, a novel supervisory structure is introduced to ameliorate the problem. The proposed structure incorporates nonlinearity compensation at the joint level (through a computer-control feedback algorithm); and performance monitoring at the task level (through discrete-time predictors. Preliminary simulation experiments demonstrate the feasibility and applicability of the approach.
|Title of host publication||Unknown Host Publication Title|
|Publisher||ISA (Robotics and Expert Systems v 3)|
|Number of pages||6|
|Publication status||Published - Dec 1 1987|
ASJC Scopus subject areas