Near-infrared spectroscopy offers a rapid and non-destructive approach for soil moisture content (SMC) prediction, yet its accuracy is constrained by heterogeneous error structures arising from soil complexity, light scattering, and instrumental noise. Current preprocessing strategies predominantly rely on empirical trial-and-error, lacking systematic analysis of error sources and their mitigation mechanisms. This study proposes an error structure-guided framework integrating error covariance matrix (ECM) and correlation matrix analysis to quantify heteroscedasticity and error coupling in raw soil spectra. Identifying dominant error types (e.g., multiplicative noise at OH absorption peaks, 1450 nm and 1940 nm) and their origins by ECM, we optimized preprocessing selection and feature band extraction. Experimental results demonstrated that multiplicative scatter correction (MSC) and standard normal variate (SNV) effectively addressed baseline shifts and heteroscedastic errors, reducing average error correlations from 0.99 to 0.20. Combined with competitive adaptive reweighted sampling (CARS), the PLS model achieved superior performance (testing set R2 = 0.99, RPD = 9.1). The proposed framework provides a universal strategy for error-aware spectral modeling, extendable to multi-parameter soil analysis (e.g., organic carbon, pH), and offers technical support for field-deployable NIR systems in precision agriculture. Future research would systematically evaluate the applicability of the framework across diverse soil matrices, including but not limited to clay and sandy loam.

The acquisition of phenotype parameters with computer vision is crucial for smart breeding, cultivation management, and automated harvesting. However, occlusion in Oudemansiella raphanipies hinders accurate segmentation and phenotype information collection. This study proposes ORP-extractor (Oudemansiella raphanipies phenotype extractor), a deep learning model designed to address the above-mentioned challenges. Initially, to realize instance segmentation of individual Oudemansiella raphanipies and acquired its complete shape, a newly improved Mask R-CNN networks (named OR R-CNN) was designed, which integrated the advantages of the Cross-Criss attention module and PointNet. Furthermore, with the shape prior of the cap-stem contour, an automatic measurement-position search method was proposed to assist in phenotype parameter extraction. Finally, four phenotypic parameters (cap diameter, cap height, stem diameter and stem length) were calculated combining the measurement positions with depth image. In addition, to increase the accuracy of annotation and save cost, a novel occlusion image synthesis strategy for ORP-extractor training also introduced. The segmentation results showed an AP of 86.58

Abstract Lighting variations, leaf occlusion, and fruit overlap make it difficult for mobile picking robots to detect and locate cucumber fruits in complex environments. This paper proposes a novel detection method based on the YOLOv4-tiny-SCE model for cucumbers in a complex environment. It combines the attention mechanism and adaptive spatial feature pyramid method to improve the detection effect of blocked and overlapping cucumbers. Additionally, the method also incorporates a loss function and clustering algorithm to enhance the accuracy and robustness of cucumber detection. On this basis, the 3D spatial coordinate model of cucumber is established using a Realsense depth camera to obtain the target image. To validate the cucumber detection and location method based on the YOLOv4-tiny-SCE model, a comparison experiment between YOLOv4-tiny-SCE and other lightweight models is conducted on the dataset. The results indicate that the YOLOv4-tiny-SCE model achieves an average detection accuracy of 99.7%. The average detection time per image is 0.006s, and there is a 2.5% increase in the F1 score. The average positioning errors of cucumber in X, Y, and Z three-dimensional space are 1.77mm, 2.9mm and 1.8 mm, respectively. This method balances target detection accuracy and model size, which is helpful in realizing the detection and location of cucumbers on low-performance airborne terminals in the future.

In order to improve the traction performance of the micro-tiller wheel on the paddy soil surface, a bionic paddy wheel was designed with a cattle hoof as the bionic prototype, and its diameter and wheel width were 0.46 m and 0.08 m, respectively. The traction performance test was carried out in a soil bin test-bed with a moisture content of 36 %. The vertical loads were 82.57 N, 131.40 N and 179.42 N, respectively. The driving speeds were 0.3 m/s, 0.5 m/s and 0.7 m/s, respectively. The drawbar pull was in the range of 10 – 120 N. The results showed that at the driving speed of 0.7 m/s, with the increase of the vertical load, the driving torque and the drawbar pull are increasing. The vertical load has a significant effect on the change of driving torque and maximum drawbar pull. Under the vertical load of 179.42 N and different driving speeds, when the slip ratio is less than 0.37, the efficiency coefficient begins to grow rapidly, and the greater the driving speed is, the greater the growth rate is. When the slip ratio is about 0.37, the efficiency coefficient reaches the maximum and then begins to decrease. Driving speed has a significant effect on the maximum efficiency coefficient of wheels. This paper can provide a reference for the traction performance of the micro-tiller wheel on the paddy soil surface and the design of the new bionic paddy wheel.