Computational adaptive optimal control with an application to blood glucose regulation in type 1 diabetics

This paper presents an online policy iteration approach for finding optimal controllers for continuous-time linear systems with completely unknown dynamics, via adaptive/approximate dynamic programming. Using the proposed scheme, the a priori knowledge of the system matrices is not required, and all iterations can be conducted by using repeatedly the same state and input information collected on some fixed finite time-intervals. A practical computational adaptive optimal control algorithm is developed in this paper, and is applied to the optimal blood glucose regulation problem in type 1 diabetics.