Aiming at the difficulty of tracking robotic fish online, this paper proposes a path planning method based on genetic algorithm. First, the grid method is used to divide the workspace of robotic fish; then an optimal robot path can be obtained in the grid environment through the genetic operations such as reproduction, crossover and mutation etc. Velocity control computation and direction control computation can be designed in terms of fuzzy control and subsection control respectively. This will make actual path and the path obtained via genetic algorithm as near as possible. Finally the computer gives corresponding control instructions to the robotic fish. Experimental results show the effectiveness of the proposed method.