Double arm robots are increasingly being used in automated production due to their higher work efficiency and better flexibility. This article focuses on the research of visual recognition based robots applied in automated production lines. Firstly, the visual system has been improved and image labeled to make it more accurate in identifying targets. Then, path planning has been carried out for the coordinated movement of both arms. Through the unconstrained collaborative planning method of both arms, the efficiency and robustness of path planning have been improved. Finally, experimental simulation has been conducted to verify the effectiveness of visual assistance conditions. The effectiveness of the dual arm planning method.