An optimal control for the convergence of a class of discrete event systems (DESs) is presented. The class of DESs represents the high-level modelling of constrained motion hybrid dynamic systems. The convergence is adapted from work published by the authors in 1996, referred to as p-convergence, and is defined in a probabilistic framework. It is shown that p-convergence is a realistic goal for practical implementation which leads to the control synthesis problem to achieve a prespecified probability of convergence. A method for discrete event controller synthesis is derived by setting this problem as a standard optimal control problem. Dynamic progamming is then used to approximate its solution. Finally, the applicability of this approach is demonstrated for a dual peg insertion task.