In this paper, reverse engineering technique was employed to extract the ridges of the hoof ball contour, and hoof ball tissue structure was analyzed based on the bionic prototype of goat hooves. The quantified geometric features were used to design the bionic track shoe pattern, which can enhance its adhesive performance and solve the problem that agricultural tracked vehicles in hilly and mountainous areas are prone to slip due to poor adhesive performance. The monolithic structure of the biomimetic goat hoof track shoe pattern and the ordinary one-line track pattern were arranged and combined; they included six kinds of track shoe models and the adhesive performance was compared. A discrete element system was established based on soil parameter determination to compare the maximum adhesion of different track shoe models. The bionic track shoe samples were prepared for soil bin tests to verify the reliability of the discrete element analysis results. Compared with the ordinary track shoe, the adhesion of the optimal bionic track shoe was improved by 9.1%.

In order to improve the slope movement stability and flexibility of quadruped robot, a theoretical design method of a flexible spine of a robot that was based on bionics was proposed. The kinematic characteristics of the spine were analyzed under different slopes with a Saanen goat as the research object. A Qualisys track manager (QTM) gait analysis system was used to obtain the trunk movement of goats under multiple slopes, and linear time normalization (LTN) was used to calibrate and match typical gait cycles to characterize the goat locomotion gait under slopes. Firstly, the spatial angle changes of cervical thoracic vertebrae, thoracolumbar vertebrae, and lumbar vertebrae were compared and analyzed under 0°, 5°, 10°, and 15° slopes, and it was found that the rigid and flexible coupling structure between the thoraco–lumbar vertebrae played an obvious role when moving on the slope. Moreover, with the increase in slope, the movement of the spine changed to the coupling movement of thoraco–lumbar coordination movement and a flexible swing of lumbar vertebrae. Then, the Gaussian mixture model (GMM) clustering algorithm was used to analyze the changes of the thoraco–lumbar vertebrae and lumbar vertebrae in different directions. Combined with anatomical knowledge, it was found that the motion of the thoraco–lumbar vertebrae and lumbar vertebrae in the goat was mainly manifested as a left–right swing in the coronal plane. Finally, on the basis of the analysis of the maximin and variation range of the thoraco–lumbar vertebrae and lumbar vertebrae in the coronal plane, it was found that the coupling motion of the thoraco–lumbar cooperative motion and flexible swing of the lumbar vertebrae at the slope of 10° had the most significant effect on the motion stability. SSE, R2, adjusted-R2, and RMSE were used as evaluation indexes, and the general equations of the spatial fitting curve of the goat spine were obtained by curve fitting of Matlab software. Finally, Origin software was used to obtain the optimal fitting spatial equations under eight movements of the goat spine with SSE and adjusted-R2 as indexes. The research will provide an idea for the bionic spine design with variable stiffness and multi-direction flexible bending, as well as a theoretical reference for the torso design of a bionic quadruped robot.

To meet rapid and non-destructive identification of selenium-enriched agricultural products selenium-enriched millet and ordinary millet were taken as objects. Image regions of interest (ROI) were selected to extract the spectral average value based on hyperspectral imaging technology. Reducing noise by the Savitzky-Golay (SG) smoothing algorithm, variables were used as inputs that were screened by successive projections algorithm (SPA), competitive adaptive reweighted sampling (CARS), uninformative variable elimination (UVE), CARS-SPA, UVE-SPA, and UVE-CARS, while sample variables were used as outputs to build support vector machine (SVM) models. The results showed that the accuracy of CARS-SPA-SVM was 100% in the training set and 99.58% in the test set equivalent to that of CARS-SVM and UVE-CARS-SVM, which was higher than that of SPA-SVM, UVE-SPA-SVM, and UVE-SVM. Therefore, the method of CARS-SPA had superiority, and CARS-SPA-SVM was suitable to identify selenium-enriched millet. Finally, 454.57 nm, 484.98 nm, 885.34 nm, and 937.1 nm, which were obtained by wavelength extraction algorithms, were considered as the sensitive wavelengths of selenium information. This study provided a reference for the identification of selenium-enriched agricultural products.

In view of the low accuracy and slow speed of goat-face recognition in real breeding environments, dairy goats were taken as the research objects, and video frames were used as the data sources. An improved YOLOv4 goat-face-recognition model was proposed to improve the detection accuracy; the original backbone network was replaced by a lightweight GhostNet feature extraction network. The pyramid network of the model was improved to a channel management mechanism with a spatial pyramid structure. The path aggregation network of the model was improved into a fusion network with residual structure in the form of double parameters, in order to improve the model’s ability to detect fine-grained features and distinguish differences between similar faces. The transfer learning pre-training weight loading method was adopted, and the detection speed, the model weight, and the mean average precision (mAP) were used as the main evaluation indicators of the network model. A total of 2522 images from 30 dairy goats were augmented, and the training set, validation set, and test set were divided according to 7:1:2. The test results of the improved YOLOv4 model showed that the mAP reached 96.7%, and the average frame rate reached 28 frames/s in the frontal face detection. Compared with the traditional YOLOv4, the mAP improved by 2.1%, and the average frame rate improved by 2 frames/s. The new model can effectively extract the facial features of dairy goats, which improves the detection accuracy and speed. In terms of profile face detection, the average detection accuracy of the improved YOLOv4 goat-face-recognition network can reach 78%. Compared with the traditional YOLOv4 model, the mAP increased by 7%, which effectively demonstrated the improved profile recognition accuracy of the model. In addition, the improved model is conducive to improving the recognition accuracy of the facial poses of goats from different angles, and provides a technical basis and reference for establishing a goat-face-recognition model in complex situations.

Accurate recognition method of pitaya in natural environment provides technical support for automatic picking. Aiming at the intricate spatial position relationship between pitaya fruits and branches, a pitaya recognition method based on improved YOLOv4 was proposed. GhostNet feature extraction network was used instead of CSPDarkNet53 as the backbone network of YOLOv4. A structure of generating a large number of feature maps through a small amount of calculation was used, and the redundant information in feature layer was obtained with lower computational cost, which can reduce the number of parameters and computation of the model. Coordinate attention was introduced to enhance the extraction of fine-grained feature of targets. An improved combinational convolution module was designed to save computing power and prevent the loss of effective features and improve the recognition accuracy. The Ghost Module was referenced in Yolo Head to improve computing speed and reduce delay. Precision, Recall, F1, AP, detection speed and weight size were selected as performance evaluation indexes of recognition model. 8800 images of pitaya fruit in different environments were used as the dataset, which were randomly divided into the training set, the validation set and the test set according to the ratio of 7:1:2. The research results show that the recognition accuracy of the improved YOLOv4 model for pitaya fruit is 99.23%. Recall, F1 and AP are 95.10%, 98% and 98.94%, respectively. The detection speed is 37.2 frames·s-1, and the weight size is 59.4MB. The improved YOLOv4 recognition algorithm can meet the requirements for the accuracy and the speed of pitaya fruit recognition in natural environment, which will ensure the rapid and accurate operation of the picking robot.