In order to combine multiple intelligent robots into a whole to complete more complex tasks, a multi robot collaborative path planning method based on cellular automata (CA) model is proposed here. The cellular automaton model is adopted to achieve the clustering and control of robotic fish, aiming to search for the best control strategies and moving paths. Moreover, with proper initial states and evolution rules, the whole process of ball topping collaborated by multiple robotic fishes is simulated. Then, the precision and the robustness of the proposed method is verified with the homemade Matlab codes, as well as the simulation platform of the International Aquatic Robot Competition. Finally, the future plan on this study is introduced and discussed.