The pattern formation of multi robot systems in different scenes has attracted more and more attention. However, the existing methods can not effectively optimize the pattern formation in obstacle environment. To solve this problem, this paper proposes a new real-time online optimization algorithm based on target matching and path optimization. Firstly, taking the distance between the robot and the virtual desired pattern as the objective function, a multi parameter pattern composition model is established, and the optimal desired pattern parameters are obtained under certain constraints. Secondly, an iterative controller is established, so that the robot can allocate the robot and the target point in real time and online in the process of moving to the target point. Then, the optimal collision avoidance speed algorithm is used to make the robot reach the target point of the desired pattern without collision, and complete the pattern formation. Finally, simulation experiments in MATLAB and V-REP are designed to verify the correctness and effectiveness of the proposed method.