Summary: Many optimal control problems have no solution because the desired final state cannot be reached by imposing an admissible control. Here we find a solution to this problem by an extension of the set of admissible controls.