End-effector Design,Seedling Information Inspection And Path Planning For Transplanting Between Vegetable Seedling Trays

Cultivation technique for seedlings in plug tray in greenhouse has been widely used in worldwide. Transplanting healthy seedlings from high-density to low-density trays for further growth, and replacing bad or missing plants with healthy ones are important tasks in cultivation. Developing automated seedling transplanter to take the place of labor to do these repetion and heavy tasks are meaningful. It will bring commercial value and socially worthwhile in China which is the great nation in facility agriculture production with vegetable varieties majority. In this paper, tomato and cucumber are mainly research objects. Four dimensions, including suitable seedlig plug status for mechanical transplanting, end-effector design and testing, methods for seedling quality evaluation based on machine vision techniques, and seedling transplanting path optimization were studied.The main research contents and conclusions were listed as follows:(1) An seedling plug clamping platform with adjustable needles component mounting on universal tester has been designed in this study which can simulate needles clamping action. Five parameters which were related with needle and plug components were tested separately to demonstrate the affection for plug compressive resistance. They were4sets of clamping needle angle with verticality (4°,7°,10°and13°),3and4gripper needles,4sets of moisture content plug (65%,75%,85%and88%),3sets of taproot length average, and2kinds of bulk material proportion. Plug got higher reliability resistance value of risen gradually in7°and10°needle angle other than4°and13°for the reason of clamping posture, especially better for7°in these4angles comparison.4needles champing can get higher resistance value than3needles’because of symmetry compression with less external deformation. Seedling plug with85%moisture content got the best performance in the4levels of moisture content plugs with good condition in material adhensionstress and plasticity. Taproot length flourish improved the shear strength and structure stability of plug. Two sets of plugs with average taproot length exceeding87mm had better resistance character than shorter taproot length plug. The proportion of bulk material which was made up of turf, roseite and perlite with6:3:1had less compression space in plug than bulk with7:2:1because of the stiff roseite dose increase. Also, the former proportion plug got a bigger resistance value with the same compression deformation. The result could provide information for end-effector design and seedling plug nursery which would be more suitable for mechanical transplanting. (2) Base on stable grasping and active separation function requirement, an plugging and clamping end-effector with two cylinder drivers was developed to do the transplanting task by movements of plugging, clamping, lifting, releasing and separation. Selection of adjustable key parameter for end-effector ensured the stable grasping for different plug dimension. And adjustment of the relative position between seedling needles and tubles realized active separation. Base on previous research of the optimal plug compressive resistance, the rate of success transplanting was100%for seedling with taproot length exceeding87mm by end-effector with3or4needles. The rate achived more than95.8%for different taproot length seedling, and end-effector with4needles had higher reliability performance than3needles’. The rate of seedling injury was smaller than3%. All of these meet the requirements of automation transplanter.The end-effector of the second-generation robotic transplanter was developed. It is driven by a single cylinder, loaded4pin or3pin fingers to meet the needs of the different crops and different specifications of the plugs. This is a fundamental solution to the problem of poor stability of two fingers actuated by two cylinders in the end-effector of the first-generation robotic transplanter and the success rate of grab has also been further improved. The result shows that this end-effector can complete the requirements of automation transplanting, the accuracy of transplanting achieved more than95%.(3) Leaf area of a seedling is an important indicator of its quality. Here, a vision system was used to measure the leaf area in each cell to distinguish "bad" and "good" plugs. Based on the principle of proportion in area, the procedures for processing top-view seedling images and a method for calculating each the leaf area of each seedling in the plug tray were investigated. Overlapping of the leaves across the surface of the cell resulted in failures in identification, which is a key point to be resolved. A decision method combining the region centre of cross-border leaves, and a methodology for the improved watershed segmentation for overlapping leaf images, were developed. Seedlings of tomato and cucumber, at suitable transplanting stages, were used to test the efficacy of the quality evaluation program. The improved watershed segmentation lessened the initial partitions by45%to55%compared with the conventional watershed algorithm. The overlapping leaves were successfully segmented. The relative identification accuracy of seedling quality was98.9%and98.6%for tomato and cucumber, respectively. The errors were mainly attributed to horticultural practices. After seedlings’further growth, the phenomenon of overlapping leaves became more serious which made the quality evaluation accuracy decline base on watershed segmentation. A method that combined image-processing procedure with mechanical separation was developed for leaf area measurement as well as to determine seedling quality for automated transplanting. A four-step image pre-processing procedure could remove the blue separators in original RGB image and extract seedlings leaves from the background. The quality identification accuracy of methods base on mechanical separator for tomato and cucumber seedlings in booming phase was100%. The method extended the period for seedling quality evaluation by machine vision technology.(4) Healthy seedlings need to be transplanted into lower density tray or vacancy cells in tray instead of unhealthy ones. The end-effector leaved the origin, and grasped the health seedlings one by one from the transplanting tray to the aim tray. And then it was back to the origin until the all vacancy cells has been filled. This process formed the automatic transplanting path. The seedling transplanting path problem was similar to traveling salesman problem which the whole path length was objective function.4regular order schemes for thin planting were optimized by greedy alogrithm, seperately. Shorter length in stable would by got by scheme3while vacancy cells were bigger than6, or emergence rate was smaller than88%; in contrast, greedy alogrithm were preferable; compared with4regular order schemes, the path planned by greedy algorithm were shortened from19.4%and4.4m to40.8%and9.55m for transplanting10trays’seedlings. A path optimization method for bad seedling vacancy cells retransplanting was developed based on genetic algorithm. The optimal result could be got in25generation’s convergence. The numerical simulated results show that compared with the fixed sequential method, the path planned by genetic algorithm has been shortened8.5%and3.7m for transplanting fifty seedlings, algorithm’s average operation time was0.65s. The transplanting path has been optimized, and operation time meets the transplanting robot’s real-time requirement.The above work provided an important basis of theoretic and research for developing automatic seedling transplanter in greenhouse which is suitable for domestic situation and industrial application in China.