To improve the precision of intelligent sugarcane harvesting, and to meet the requirements of high precision and low complexity for use with embedded devices, a lightweight model called MobileNet v2-YOLO v5s for the real-time detection of sugarcane stem nodes in complex natural environments was developed. In this study, images of sugarcane stem nodes in a complex natural environment were collected and a dataset containing 12,600 images was constructed using a data extension process. The MobileNet network was introduced to replace the backbone of the YOLO v5s algorithm and the improved algorithm was used to train the MobileNet-YOLO v5s sugarcane stem node identification model. In experiments aiming to verify the advantages of the lightweight model, MobileNet v2-YOLO v5s achieved the best combination of high precision and low complexity. Its AP was decreased by only 0.8%, while its complexity was reduced by 40% compared to YOLO v5s. It also had a fast detection speed of 4.4 ms on a Dell workstation P7920. Therefore, 11 other models were selected for comparative experiments to demonstrate the superiority of MobileNet v2-YOLO v5s. Finally, TensorRT accelerated optimization tests, execution tests, and real-time detection tests were performed on Jetson Nano. The results showed that the optimised MobileNet v2-YOLO v5s outperformed YOLO v5s in terms of identification, lightweight and detection speed on embedded devices. Overall, MobileNet v2-YOLO v5s model meets the requirements of embedded devices and can provide a visual identification method for intelligent sugarcane harvesting.

The rapid and accurate identification of sugarcane stem nodes in the complex natural environment is essential for the development of intelligent sugarcane harvesters. However, traditional sugarcane stem node recognition has been mainly based on image processing and recognition technology, where the recognition accuracy is low in a complex natural environment. In this paper, an object detection algorithm based on deep learning was proposed for sugarcane stem node recognition in a complex natural environment, and the robustness and generalisation ability of the algorithm were improved by the dataset expansion method to simulate different illumination conditions. The impact of the data expansion and lighting condition in different time periods on the results of sugarcane stem nodes detection was discussed, and the superiority of YOLO v4, which performed best in the experiment, was verified by comparing it with four different deep learning algorithms, namely Faster R-CNN, SSD300, RetinaNet and YOLO v3. The comparison results showed that the AP (average precision) of the sugarcane stem nodes detected by YOLO v4 was 95.17%, which was higher than that of the other four algorithms (78.87%, 88.98%, 90.88% and 92.69%, respectively). Meanwhile, the detection speed of the YOLO v4 method was 69 f/s and exceeded the requirement of a real-time detection speed of 30 f/s. The research shows that it is a feasible method for real-time detection of sugarcane stem nodes in a complex natural environment. This research provides visual technical support for the development of intelligent sugarcane harvesters.

Spatial location of sugarcane nodes using robots in agricultural conditions is a challenge in modern precision agriculture owing to the complex form of the sugarcane node when wrapped with leaves and the high computational demand. To solve these problems, a new binocular location method based on the improved YOLOv4 was proposed in this paper. First, the YOLOv4 deep learning algorithm was improved by the Channel Pruning Technology in network slimming, so as to ensure the high recognition accuracy of the deep learning algorithm and to facilitate transplantation to embedded chips. Secondly, the SIFT feature points were optimised by the RANSAC algorithm and epipolar constraint, which greatly reduced the mismatching problem caused by the similarity between stem nodes and sugarcane leaves. Finally, by using the optimised matching point to solve the homography transformation matrix, the space location of the sugarcane nodes was for the first time applied to the embedded chip in the complex field environment. The experimental results showed that the improved YOLOv4 algorithm reduced the model size, parameters and FLOPs by about 89.1%, while the average precision (AP) of stem node identification only dropped by 0.1% (from 94.5% to 94.4%). Compared with other deep learning algorithms, the improved YOLOv4 algorithm also has great advantages. Specifically, the improved algorithm was 1.3% and 0.3% higher than SSD and YOLOv3 in average precision (AP). In terms of parameters, FLOPs and model size, the improved YOLOv4 algorithm was only about 1/3 of SSD and 1/10 of YOLOv3. At the same time, the average locational error of the stem node in the Z direction was only 1.88 mm, which totally meets the demand of sugarcane harvesting robots in the next stage.

Carbon fiber-reinforced polymer (CFRP) drilling is a typical process in the aircraft industry. Because the components of CFRP are different and uneven, it is difficult to extract tool wear characteristics from the machining signals, which are composed of the processing characteristics of various materials and the tool state characteristics. The aim of this work is to present a new comprehensive approach based on multicharacteristics and multisignal sources to predict the tool wear state during CFRP drilling through a combination of a backpropagation (BP) artificial neural network (ANN) model and an efficient automatic system depending on the sliding window algorithm. It was verified that the peak factor and Kurtosis coefficient of different signals and the energy value of the d5 layer of the thrust force signal and the d3 layer of the vibration signal after wavelet decomposition were related to tool wear. Among them, the energy value of the d3 layer of the vibration signal was selected as the wear indicator and was able to describe the state of the tool during the CFRP drilling process regardless of the drilling conditions and individual tool differences. A confirmatory drilling experiment using 6-mm-diameter polycrystalline diamond twist drilling under different processing parameters was conducted to verify the ANN model based on multicharacteristics and multisignal sources. A lower feed speed and a higher cutting speed were both highly correlated with the VB value of flank wear. Drill wear accelerated because of the occurrence of adhesive wear when the number of drilled holes reached around 90. The accuracy of the neural network model is 80–87% when using the value of only one characteristic but clearly increases based on multicharacteristics and multisignal sources in real time, indicating that the BP ANN model has higher accuracy in predicting the tool state in CFRP drilling through the sensor signal fusion method.