Seedlings need to be transplanted from high density into lower density trays for more space to grow in the greenhouse. Automatic transplanter could detect seedlings health state and do the tasks of lower density transplanting. It can work in high efficiency and good quality comparing with the traditional manual work. The lower density transplanting path includes endeffector leaving from original point, grasping health seedlings from high density to lower density trays one by one, and going back to the start at last. The distances of transplanting path were decided by seedlings grasping order. Traversing search algorithm consumed large calculation to plan this transplanting path which does not meet the real time requirement. In this paper, four schemes of fixed sequences were optimized by using the greedy algorithm separately, and eight schemes were formed totally. The path planning methods were used in the sparse and dense trays to verify the effectiveness of the greedy algorithm scheme. Comparing with the longest mean value of fixed sequence schemes for high density trays path planning, optimal scheme of GAS3 could get a better result with more than 10.6% amplitude optimization. The average calculation time of the optimization scheme was 0.84s. Significance test showed the increase of vacancy holes reduced the effectiveness of length shorten. As a whole, the greedy algorithm scheme optimizes the lower density transplanting path, meets the realtime work requirement, and improves the transplanting work efficiency.