An algorithm for constructing of optimal strategy in a nonlinear differential game with a non-fixed completion time

We develop a method to compute quasioptimal strategies in a nonlinear differential game on a non-fixed time interval with a goal set. In two-dimensional case the play-attainable sets are calculated by an algorithm, similar to one for convolution of Minkowskis sum of two polyhedra. We provide detail error estimations of the algorithm.