At this stage, safflower picking is mostly performed manually or semi-manually, the picking method is antiquated and the picking precision is low. In this experimental study, a new attitude tilt levelling system was designed for a safflower-picking robot, which has created a solid foundation for the realization of future safflower-picking machine automation. The mobile platform was simplified as a four-point support, and an automatic levelling control system was designed based on the multi-sensor data collected by a multi-inclination sensor, a multi-pressure sensor, and a displacement sensor. The error range of the levelling of the mobile platform was obtained by MATLAB simulation analysis, the relationship between the inclination of the mobile platform and the displacement of the levelling mechanism was analyzed by coordinate transformation, and the maximum levelling range of the levelling mechanism was analyzed. On this basis, an automatic levelling control system was designed. Finally, the safflower-picking mobile platform was tested, and we concluded that the levelling control system can adjust the inclination angle of the mobile platform to within 0.2° and the levelling time to within 7 s. The design of the automatic levelling control system fills the gap in the field of safflower picking and adopts multi-sensor fusion. Compared with other methods, the collected inclination data is more accurate, the levelling accuracy higher, and the levelling time shorter. The final results show that this experimental study provides a strong basis for the realization of the full-mechanical automation of safflower picking.

In response to the challenge of low detection and positioning accuracy for safflower corollas during field operations, we propose a deep learning-based object detection and positioning algorithm called the Mobile Safflower Detection and Position Network (MSDP-Net). This approach is designed to overcome issues related to the small size of safflower corollas and their tendency to be occluded in complex agricultural environments. For object detection, we introduce an improved YOLO v5m model, referred to as C-YOLO v5m, which integrates a Convolutional Block Attention Module (CBAM) into both the backbone and neck networks. This modification enhances the model’s ability to focus on key features, resulting in increases in the precision, recall, and mean average precision of 4.98%, 4.3%, and 5.5%, respectively. For spatial positioning, we propose a mobile camera-based method in which a binocular camera is mounted on a translation stage, enabling horizontal movement that maintains optimal positioning accuracy and mitigates occlusion issues. Field experiments demonstrate that this mobile positioning method achieves a success rate of 93.79% with average deviations of less than 3 mm in the X, Y, and Z directions. Moreover, comparisons with five mainstream object detection algorithms reveal that MSDP-Net offers superior overall performance, making it highly suitable for safflower corolla detection. Finally, when applied to our self-developed safflower harvesting robot, 500 indoor trial tests achieved a harvest success rate of 90.20%, and field tests along a 15 m row confirmed a success rate above 90%, thereby validating the effectiveness of the proposed methods.

This study proposed a standardized oilseed rape seedling-carrying potting molding method to improve the adaptability of mechanical transplanting of potting seedlings. This method aims to address the failure in seedling pick-up and transport during the mechanized transplanting of rapeseed pot seedlings, which is caused by matrix breakage and seedling damage. This study selected cylindrical oilseed rape seedling-carrying potting as the research object and investigated the relationship between the physical characteristics of seedling-carrying potting and the proportion of the composition of the matrix soil as well as the characteristics of seedling growth after planting. The optimal parameter combination of the matrix soil was obtained using Design-Expert 8.0.6 software: dry matter ratio of 4:1, compression ratio of 0.36, and moisture content of 45%. A single-factor test was conducted using a seedling-carrying potting test bed. According to the single-factor test results, the dry matter ratios (commercial substrate: clay loam mass ratios of 2:1, 3:1, and 4:1), matrix soil compression ratios (0.35, 0.40, and 0.45), and matrix soil moisture content (35%, 40%, and 45%) were selected as the factors of influence, while the drop loss rate, shear resistance, and scattering rate were used as the indicators of evaluation. The drop loss rate of seedling-carrying potting under this parameter combination was 1.5%, the shear resistance was 7.1 N, and the scattering rate was 34.9%. Validation tests were conducted on a seedling-carrying potting test bed, and the relative errors between the actual and simulated values of the drop loss rate, shear resistance, and scattering rate were 7.1%, 7.0%, and 8.4%, respectively, verifying the accuracy of the model and the optimized parameters. Comparison tests of the growth characteristics of the optimized seedling-carrying potting, hole-tray seedling, and bare seedling in field transplanting were conducted. The results displayed that root length, root diameter, root dry matter, chlorophyll content, and seedling vigor index consistently followed the same descending order: seedling-carrying potting > hole-tray seedlings > bare seedlings. Compared to hole-tray seedlings, the corresponding growth characteristics of seedling-carrying potting were 11.7%, 10%, 21.7%, 2.8%, and 27.8% higher, respectively. Compared to bare seedlings, they were 17.1%, 12.5%, 32.2%, 10.8%, and 32.7% higher, respectively. The seedling length, seedling width, plant taper angle, and dry matter mass of stem and leaves were, in descending order, greater in hole-tray seedlings, followed by seedling-carrying potting, and then bare seedlings. In comparison, the corresponding growth characteristics of seedling-carrying potting were 8.9%, 9.8%, 2.3%, and 30.6% higher than those of bare seedlings, respectively.