Explicit sub-optimal linear quadratic regulation with state and input constraints (Q1614350)

A suboptimal strategy for explicit off-line design of linear-quadratic controllers subject to state and input constraints is derived. The Hamilton-Jacobi-Bellmann (HJB) equation of the problem is decomposed into two nested parts. The inner part of the HJB equation involves only equality constraints. Its solution is an affine state feedback. The outer part of the HJB equation is a discrete optimization problem over all allowed active constraint set sequences which can be also solved off-line or reduced to a simpler problem and then solved efficiently in real-time. It is demonstrated that the proper choice of a finite horizon will lead to the significant reduction in the complexity of the state-space partitioning. Some other aspects of suboptimality, computational complexity and real-time implementation are also discussed.