Computing policies and performance bounds for deterministic dynamic programs using mixed integer programming

We present a method for computing control policies and performance bounds for deterministic finite-state/finite-action dynamic programs, with the infinite horizon discounted cost, using a receding horizon approach. The computations are done by solving mixed integer programs of moderate size. The efficacy of the proposed technique is demonstrated on a job-shop scheduling example.

Citation

Cogill, R.; Hindi, H. Computing policies and performance bounds for deterministic dynamic programs using mixed integer programming. American Control Conference (ACC); 2011 June 29 - July 1; San Francisco CA.