A structured robust optimal control technique

Optimal control provides a systematic procedure for calculating nominal trajectories and control functions, and for taking account of high-level objectives of control, embodied in a cost function. This paper proposes including a sensitivity term in the cost, scaled by a parameter β, whose purpose is to de-sensitize key variables to plant variation. Adjusting β permits a trade-off to be achieved between sensitivity and cost degradation. We show that the method compares favorably with other de-sensitization techniques.