Chunfeng Gao

To address the difficulty of stable axial orientation determination of cucurbit rootstock seedlings under random initial postures, true-leaf interference, and mechanical occlusion during mechanized grafting, a two-stage axial orientation recognition and rotational correction method integrating mechanical constraint and machine vision was proposed. In the first stage, seedling posture was normalized using the PCA axis of the whole green region, and the seedling was rotated to avoid the leaf hold-down ring handle. In the second stage, mechanically occluded leaf regions were restored after ring constraint, and a minimum-span half-area window method based on polar-angle distribution was proposed to determine the directed cotyledon axis. The method was compared with the rotation-center–cotyledon-centroid line method and the single-cotyledon PCA method. Fifty seedlings were tested at each of the true-leaf emergence, initial true-leaf expansion, and true-leaf extension stages. In 150 first-stage rotation trials conducted on 50 seedlings, the absolute PCA-axis deviation after rotation was within 1.5° in every trial. In a separate handle-avoidance test, the avoidance rate was 100%. In the second stage, the proposed method yielded median absolute errors of 0.71°–1.26° for both cotyledons, with no significant differences among growth stages. In contrast, the errors of the centroid-line and PCA methods increased significantly with seedling development, reaching maximum median errors of 5.98° and 19.22°, respectively. After the second-stage rotation, the algorithm-derived residual axial deviation of the right cotyledon was below 1° for all 150 seedlings. These results indicate that the proposed method provides stable cotyledon-axis recognition under the investigated experimental conditions.

Seed dimensions (length and width), surface color, frictional properties (static and kinetic coefficients of friction), thousand-seed weight, and angle of repose are five important categories of seed physical properties. Conventional methods generally measure these properties separately and rely heavily on manual operation, resulting in limited applicability and difficulty in balancing measurement efficiency and accuracy. To address these limitations, this study developed an automated method and an integrated platform incorporating automatic feeding, individual-seed positioning, state recognition, parameter acquisition, and cyclic control. After a single sample loading, the platform sequentially processed individual seeds, measured their dimensions, surface color, and static and kinetic coefficients of friction, and accumulated the measured seeds for subsequent thousand-seed weight and angle-of-repose determination, thereby enabling continuous automated measurement of the five categories of physical properties. Experiments were conducted using maize kernels, red kidney beans, and sunflower seeds to evaluate the measurement performance, repeatability, and cross-material adaptability of the platform. The standard deviations of repeated seed-dimension measurements were below 0.042 mm for all three seed types. The first moments of H and S in the HSV color space and a* and b* in the CIELAB color space were selected for color analysis and exhibited relatively low sensitivity to illumination variation under the tested imaging conditions. Within the stable-sliding intervals, the coefficients of determination for the relationship between actual seed displacement and squared time were all greater than 0.99, supporting the approximation of uniformly accelerated translational motion and the calculation of sliding acceleration for kinetic-friction determination. When the sample size used for thousand-seed weight estimation exceeded 50 seeds, further reductions in the coefficient of variation and average relative error were no greater than 0.2 percentage points. The angle-of-repose measurements distinguished differences in the pile formation characteristics of the three seed types. All three seed types completed the entire measurement procedure, corresponding to a platform adaptation success rate of 100%. In the cross-material validation, the coefficients of variation of all evaluated repeated-measurement indicators were no greater than 2.5%. These results indicate that, under the tested material and experimental conditions, the platform exhibited good operational stability, measurement repeatability, and cross-material adaptability, providing an effective approach for the automated and integrated measurement of multiple seed physical properties.

In the deep-water lifting system, the variation of the lifted load motion and cable tension increases dramatically with the heave motions of the installation ship in the ocean. To compensate for the heavy velocity changes of the lifted load, the heave compensation controller is designed by using the model-free adaptive control approach, which is based on compact form dynamic linearization to achieve higher compensation precision. The feedforward prediction controller is designed by adopting time series analysis to solve the time-delay problem of this compensation system. The servo valve controlling the hydraulic motor is designed by a hydraulic driving system, and a differential planetary gear winch is used as the transmission. The simulation model is built using the transfer function of the system. The simulation study demonstrates that the model-free-adaptive-control scheme applied to the deep-water installation heave-compensation system has excellent dynamic performance and robustness, with a compensation rate of above 80%. In order to confirm the control algorithm and verify the possibility of the active control system, tests are executed in a laboratory room.