The detection of obstacles of intelligent driving vehicles becomes a primary condition to ensure safe driving. LiDAR and camera are the main sensors for intelligent vehicles to obtain information about their surroundings, and they each have their own benefits in terms of object detection. LiDAR can obtain the position and geometric structure of the object, and camera is very suitable for object recognition, but the reliance on environmental perception by a single-type sensor can no longer meet the detection requirements in complex traffic scenes. Therefore, this paper proposes an improved AVOD fusion algorithm for LiDAR and machine vision sensors. The traditional NMS (non-maximum suppression) algorithm is optimized using a Gaussian weighting method, while the 3D-IoU pose estimation loss function is introduced into the target frame screening module to upgrade the 2D loss function to 3D and design the 3D-IoU criterion. By comparing the detection accuracy of the algorithm proposed in this paper with that of the traditional method, it has been found that the improved AVOD fusion algorithm significantly improved the detection efficiency and the detection accuracy reached 96.3%. The algorithm proposed in this paper can provide a new approach for object detection of intelligent driving vehicles.

In view of the difficulties in weeding and plant protection in the middle and late period of maize planting, this paper proposed a self-propelled thermal fogger chassis. According to the theoretical calculation and agronomic requirements for maize planting, the structure and working principles of the self-propelled thermal fogger chassis were introduced. On this basis, the multi-body dynamics model of chassis structure was established, and the chassis traction, steering and obstacle surmounting performances were also analyzed. Then the rationality and the feasibility of the design were verified through the furrow running test and test equipped with thermal fogger. Test results showed that, the traction performance improves with the decrease of soil deformation index and increase of cohesion, and when track pre-tensioning force was about 1000 N, the machine had a good traction performance; with the decrease of the soil deformation index and the increase of cohesive force, the stability of the single side brake turn of the chassis becomes better; on the contrary, with the increase of the tightness of the crawler, the steering radius turns smaller and the steering stability becomes worse. Under heavy clay, with the pre-tensioning of 1000 N, the machine has better steering stability and smaller turning radius. The obstacle-surmounting simulation result shows that on sandy soil road, the maximum climbing angle for the chassis is 42°, the height of vertical obstacle crossing is 170 mm and the trench width is 440 mm. The study provides a reference for the design of plant protection machinery in the middle and late stages of maize planting. Keywords: maize inter-rows, intelligent chassis, traction performance, steering performance, obstacle surmounting performance DOI: 10.25165/j.ijabe.20181105.3607 Citation: Chen L Q, Wang P P, Zhang P, Zheng Q, He J, Wang Q J. Performance analysis and test of a maize inter-row self-propelled thermal fogger chassis. Int J Agric & Biol Eng, 2018; 11(5): 100–107.

Soil block distribution is one of the important indexes to evaluate the tillage performance of agricultural machinery. The traditional manual screening methods have the problems of low efficiency and damaging the original surface of the soil. This study proposes a statistical method of farmland soil block distribution based on deep learning. This method combines the adaptive learning rate and squeeze-and-excitation networks channel attention mechanism based on the original Mask-RCNN and uses the improved model to identify, segment and distribute statistics of the farmland soil blocks. Firstly, the influence of different learning rates and an improved Mask-RCNN algorithm model on training results were analyzed. Secondly, the effectiveness of the model in soil block identification and size measurement was analyzed. Finally, the identified soil blocks were classified accordingly, and the scale problem of soil block distribution after removing edge soil blocks was analyzed. The results show that with the decrease of learning rate, the loss value of model training decreases and the prediction accuracy of model is improved. The average precision value of the improved model increased by 25.29 %, and the recall value increased by 8.92%. The correlation coefficient of the maximum diameter measured by manual measurement and the maximum diameter measured by model algorithm was 0.99, which verifies the feasibility of the algorithm model. The prediction error of the model is the smallest when the camera height is 40 cm. Large-scale detection of soil block size in an experimental field in Hefei, Anhui, with an average confidence of over 97%. At the same time, the soil block is effectively classified according to the set classification standard. This study can provide an effective method for the accurate classification of soil block size and can provide a quantitative basis for the control of farmland cultivation intensity.

Traditional water–fertilizer control systems often suffer from poor precision and slow response, limiting precision agriculture development. This study developed an electrical conductivity (EC) control system for water–fertilizer integration using a fuzzy Proportional-Integral-Derivative (PID) controller optimized by particle swarm optimization (PSO) and integrated with IoT technology. MATLAB/Simulink simulations showed the proposed controller achieved the smallest overshoot (7.64–8.15%), with average settling time reduced by 62.48 s and 20.38 s compared to conventional PID and fuzzy PID controllers, respectively (p < 0.001). Field experiments on winter wheat demonstrated a mean absolute EC deviation of 0.01125 ms/cm, with root-mean-square error (RMSE) of 0.0217 ms/cm, indicating high precision under field conditions. The system also maintained soil moisture in the optimal range (19–25%) with high irrigation uniformity (Christiansen’s coefficient Cu = 97.6%). The system maintained soil moisture in the optimal range (19–25%) while supporting stable soil nutrient levels and crop growth parameters. This study provides a validated technical solution for precision EC control while establishing a foundation for future fully integrated water–fertilizer management systems.

The identification of seedling numbers is directly related to the acquisition of seedling information, such as survival rate and emergence rate. It indirectly affects detection efficiency and yield evaluation. Manual counting methods are time-consuming and laborious, and the accuracy is not high in complex backgrounds or high-density environments. It is challenging to achieve improved results using traditional target detection methods and improved methods. Therefore, this paper adopted the density estimation method and improved the population density counting network to obtain the rapeseed seedling counting network named BCNet. BCNet uses spatial attention and channel attention modules and enhances feature information and concatenation to improve the expressiveness of the entire feature map. In addition, BCNet uses a 1 × 1 convolutional layer for additional feature extraction and introduces the torch.abs function at the network output port. In this study, distribution experiments and seedling prediction were conducted. The results indicate that BCNet exhibits the smallest counting error compared to the CSRNet and the Bayesian algorithm. The MAE and MSE reach 3.40 and 4.99, respectively, with the highest counting accuracy. The distribution experiment and seedling prediction showed that, compared with the other density maps, the density response points corresponding to the characteristics of the seedling region were more prominent. The predicted number of the BCNet algorithm was closer to the actual number, verifying the feasibility of the improved method. This could provide a reference for the identification and counting of rapeseed seedlings.

The passing performance and driving stability performance of HGPM in an unstructured environment will directly affect the quality of HGPM operation. This paper designs an HGPM power chassis transmission system in order to address the problem of poor adaptability of existing plant protection machines to complex working conditions in the field, especially in the middle and late stage of plant protection operation of tall crops, which cannot pass smoothly due to the obstruction of vision and special road conditions resulting in insufficient traction of the whole machine. The system is theoretically analyzed based on hydrostatic transmission and a time-sharing four-wheel drive splitter; then, based on Solidworks and RecurDyn software, the HGPM is modeled in three dimensions, and the dynamic simulation of working conditions such as climbing, crossing the ridge, and opening the road during field operation is carried out. The simulation results show that the 2H mode can climb over a slope with an angle in the range of 0−25° and a ridge with height in the range of 0−100 mm, the 4H mode can climb over a slope with an angle in the range of 0−35° and a ridge with height in the range of 0−320 mm, with relatively stable body speed and the wheel rotation angular speed converging faster under the open road condition. Finally, prototype performance tests were conducted. The test results show that 4H mode can smoothly pass the ridge with a ridge height of 320 mm and a slope of 26°, while 2H mode has a sharp drop in speed to 0 after a short fluctuation. 4H mode achieves a more rapid convergence of longitudinal wheel stability compared to 2H mode. The developed chassis drive system of a new type of HGPM meets the design requirements and provides a reference for the dynamic chassis design of HGPM.