An improved YOLOv5 algorithm for the efficient recognition and detection of asparagus with a high accuracy in complex environments was proposed in this study to realize the intelligent machine harvesting of green asparagus. The coordinate attention (CA) mechanism was added to the backbone feature extraction network, which focused more attention on the growth characteristics of asparagus. In the neck part of the algorithm, PANet was replaced with BiFPN, which enhanced the feature propagation and reuse. At the same time, a dataset of asparagus in complex environments under different weather conditions was constructed, and the performance variations of the models with distinct attention mechanisms and feature fusion networks were compared through experiments. Experimental results showed that the mAP_@0.5 of the improved YOLOv5 model increased by 4.22% and reached 98.69%, compared with the YOLOv5 prototype network. Thus, the improved YOLOv5 algorithm can effectively detect asparagus and provide technical support for intelligent machine harvesting of asparagus in different weather conditions and complex environments.

This study is concerned with path planning in a structured greenhouse, in contrast to much of the previous research addressing applications in outdoor fields. The prototype mainly comprises an independently driven Mecanum wheel, a lidar measuring module, a single-chip microcomputer control board, and a laptop computer. Environmental information collection and mapping were completed on the basis of lidar and laptop computer connection. The path planning algorithm used in this paper expanded the 8-search-neighborhood of the traditional A* algorithm to a 48-search-neighborhood, increasing the search direction and improving the efficiency of path planning. The Floyd algorithm was integrated to smooth the planned path and reduced the turning points in the path. In this way, the problems of the traditional A* algorithm could be solved (i.e., slow the path planning speed and high numbers of redundant points). Tests showed that the turning points, planning path time, and distance of the improved algorithm were the lowest. Compared with the traditional 8-search-neighborhood A* algorithm, the turning point was reduced by 50%, the planning time was reduced by 13.53%, and the planning distance was reduced by 13.96%. Therefore, the improved method of the A* algorithm proposed in this paper improves the precision of the planning path and reduces the planning time, providing a theoretical basis for the navigation, inspection, and standardization construction of greenhouses in the future.

Mature broccoli has large flower balls and thick stems. Therefore, manual broccoli picking is laborious and energy-consuming. However, the big spheroid vegetable-picking manipulator has a complex structure and poor enveloping effect and easily causes mechanical damage. Therefore, a broccoli flower ball-picking manipulator with a compact structure and simple control system was designed. The manipulator was smart in structure and stable in configuration when enveloped in flower balls. First, a physical damage test was carried out on broccoli according to the underactuated manipulator’s design scheme. The maximum surface pressure of the flower ball was 30 N, and the maximum cutting force of the stem was 35 N. Then, kinematic analysis was completed, and the statical model of the underactuated mechanism was established. The dimension of the underactuated mechanism for each connecting rod was determined based on the damage test results and design requirements. The sizes of each connecting rod were 50 cm, 90 cm, 50 cm, 90 cm, 50 cm, 60 cm, and 65 cm. The statical model calculated the required thrust of the underactuated mechanism as 598.66–702.88 N. Then, the manipulator was simulated to verify its reliability of the manipulator. Finally, the manipulator’s motion track, speed, and motor speed were determined in advance in the laboratory environment. One-hundred picking tests were carried out on mature broccoli with a 135–185 mm diameter. Results showed that the manipulator had an 84% success rate in picking and a 100% lossless rate. The fastest single harvest time in the test stand was 11.37 s when the speed of the robot arm was 3.4 m/s, and the speed of the stepper motor was 60 r/min.

To solve the design problem of non-circular gears with cusp pitch curves, this paper proposed a new variable-involute and incomplete variable-cycloid composite tooth profile (VIIVC-CTF), deduced the new VIIVC-CTF mathematical model, and constructed the conjugate gear model based on the envelope method. The design software of the non-circular gear with a cusp pitch curve was developed based on MATLAB. The variation law of rolling radius on an incomplete cycloid profile and its characteristics such as pressure angle and radius of curvature were analyzed. The variation relationship of the rolling radius on the meshing line and the contact ratio of the VIIVC-CTF were studied. The variation relationship of incomplete variable-cycloid profile shape, pressure angle, and curvature radius corresponding to different elliptical eccentricities were analyzed. The meshing analysis of the non-circular gear transmission mechanism was carried out based on virtual software. A comparison of the consistency of the theoretical value and simulation value of the transmission ratio curve verified that the tooth profile design method was feasible, and the VIIVC-CTF was applied to the seedling pick-up mechanism of the non-circular gear planetary gear train. Through the seedling picking experiment of the seedling pick-up mechanism, the feasibility of the application of the VIIVC-CTF was verified.