Replugging tasks make seedling in well consistency in greenhouses. Healthy seedlings are used to replace the ungerminated or poor growth seedlings. This task is labor intensive by traditional manual method. And automated transplanters do the replugging task in high efficiency and good quality. According to the seedlings healthy information which is detected by machine vision, end-effector grasping healthy seedlings does the repetitive replugging task. The position of vacancy holes in plug tray are randomly. Optimizing the seedling grasping sequence can decrease the transplanting path which can improve working efficiency. A greedy genetic algorithm (GGA) was proposed for replugging tour planning which combined the character of greedy algorithm (GAS) and genetic algorithm (GA). The algorithm was robustness. The GGA was suitable for sparse and dense trays’ path optimization when segmentation step value and hereditary algebra were 8 and 100, respectively. The average path deviation of GGA and GA was 443 mm. And their effectiveness was better than that of GAS. Compared with fixed sequence method (FS), the range of optimization amplitude for GGA was 33.8%～41.3%. GA and GGA could finish the optimization operation in 1.81s and 5.59s, respectively. The results showed that GGA was more suitable for the action requirement between delivery unit and transplanting unit. The working efficiency of automated transplanter was further improved.