In the game, we generally use A* algorithm for pathfinding according to the target location and map information to realize the tracking and interception task. This approach is simple to implement, but is often less than ideal in its presentation, because it does not allow game units to predictably choose paths based on dynamic targets or obstacles, and may lead to blocked channels or difficulty in choosing the actual optimal path. This paper focuses on analyzing the need of a unit when intercepting another unit in the game, and make an interception scheme which can adapt itself to dynamic obstacles and moving target thus presents artificial intelligence to some extent.