Obstacle avoidance flight is an important problem to be solved in cluster formation flight of UAV in complex battlefield environment. In order to achieve efficient and reliable obstacle avoidance flight, based on the distributed leader-following consensus control strategy, the UAV cluster is divided into the leader in charge of path planning and the follower responsible for formation control. An improved RRT* path planning algorithm is proposed to provide real-time path planning for the UAV cluster in complex environments with dynamic obstacles. Based on the backstepping control method and taking the formation consensus protocol as the error, a distributed formation flight control law is developed to dynamically adjust the formation spacing according to the environment. By constructing three kinds of obstacle environments with different complexity to validate the formation obstacle avoidance algorithm, simulation results show that the proposed approach may generate feasible formation flight paths rapidly corresponding to the environment changes, and realize the accurate control of variable spacing formation flight of UAV cluster, which achieves the goal of efficient obstacle avoidance for the cluster formation flight.