Interior point methods in the design of hierarchical controllers

This paper reports on the use of interior point (IP) optimization algorithms for determining an “optimal” set of control agents in a single layer of a hierarchical control system. The proposed algorithm is a modification of a primal IP algorithm which is closely related to alternating minimization techniques employed by neural network researchers. Preliminary simulation and analysis results suggest that the proposed algorithm has a computational complexity of O(n^3.5L) where n and L parameterize the control problem´s size and accuracy, respectively