Body tilt is typically unavoidable for agricultural vehicles or platforms when operated over tough terrain in agricultural production, and, therefore, it is not possible to guarantee the operation performance. Current leveling methods mainly focus on wheeled vehicles or adjustment for single working parts rather than the vehicle body, and few of them could be applied to crawler vehicles. The objective of this research was to put forward an adaptive leveling system for combine harvesters based on the development of a new four-point lifting adjustable crawler chassis. The working principle of the proposed adjustable mechanism was studied on the basis of the analysis of adjustment characteristics under multiple posture adjustment conditions in a RecurDyn environment. The design relies on the combination of the attitude detection of the leveling system and adjustment calculation of driving hydraulic cylinders according to the established mathematical models. Within the designed adjustable extent (including adjustment range of vehicle height 0–87.8 mm, lateral inclination ±3.98° and longitudinal inclination −2.9–5.2°), the leveling performance of lateral and longitudinal inclination was then tested separately. The experiment showed that the leveling system could achieve automatic leveling with an accuracy of ±0.4° and could provide technical support for the development of crawler vehicles.

The combine harvester equipped with attitude-adjustment functionality significantly enhances its adaptability to complex terrain but often struggles to maintain the reliability of its mechanisms. Therefore, investigating the dynamic load characteristics of the attitude-adjustment mechanism becomes imperative. This article employed the DEM–FMBD (Discrete Element Method–Flexible Multibody Dynamics) bidirectional coupling simulation method to establish a multibody dynamic model of a tracked combine harvester. The study delved into the interaction mechanism and dynamic stress response characteristics between the tracked chassis and the complex terrain under various height adjustments, lateral adjustment angles, longitudinal adjustment angles, and different field-ridge crossing methods. Finally, the accuracy of the coupled simulation model was validated through a constructed stress detection system. The research findings revealed that the displacement and tilt angle deviation of the hydraulic cylinders utilized to execute the chassis adjustment actions in the constructed coupled simulation model was less than 5%, and the deviation between the simulation results and the actual maximum dynamic stress under multiple working conditions ranged from 7% to 15%. This verification confirmed the effectiveness of the DEM–FMBD coupled simulation method. Under different adjustment conditions, the maximum stress position was consistently distributed in the same area of the left-front and left-rear rotating arms. The primary and secondary effects of the various parts of the adjustment mechanism on the overall reliability of the chassis were as follows: left front > right front > left rear > right rear. By implementing the middle height with the adjustment strategy, the dynamic stress extreme value of the adjustment mechanism can be effectively reduced by 21.98%, thereby enhancing the structural stability of the chassis.

Aimed at addressing the problems of the existing straw choppers on combine harvesters, such as a large cutting resistance and poor cutting effect, combined with bionic engineering technology and biological characteristics, a bionic model was used to extract the characteristics of the cutting blades of locusta migratoria manilensis’s upper jaw. A 3D point cloud reconstruction and machine vision methods were used to fit the polynomial curve of the blade edge using Matlab 2016. A straw-cutting process was simulated using the discrete element method, and the cutting effect of the bionic blade was verified. Cutting experiments with rice straws were conducted using a physical property tester, and the cutting resistance of straw to bionic blades and general blades was compared. On the whole, the average cutting force of the bionic blades was lower than that of the general blades. The average cutting force of the bionic blade was 18.74~38.23% lower than that of a smooth blade and 1.63~25.23% lower than that of a serrated blade. Similarly, the maximum instantaneous cutting force of the bionic blade was reduced by 2.30~2.89% compared with the general blade, which had a significant drag reduction effect. By comparing the time–force curves of different blades’ cutting processes, it was determined that the drag-reducing effect of the bionic blade lies in shortening the straw rupture time. The larger the contact area between the blade and the straw, the more uniform the cutting morphology of the straw after cutting. Field experiment results indicate that the average power consumption of a straw chopper partially installed with bionic blades was 5.48% lower than one with smooth blades, measured using a wireless torque analysis module. In this research study, the structure of the straw chopper of an existing combine harvester was improved based on the bionic principle, which reduced resistance when cutting crop straw, thus reducing the power consumption required by the straw chopper and improving the effectiveness and stability of the blades.

The precise, real-time delineation of rice lodging areas constitutes a fundamental prerequisite for the adaptive operation of unmanned combine harvesters. However, existing deep learning methods struggle to resolve a critical limitation: achieving an optimal equilibrium between robust regional morphological perception—which is crucial for irregular lodging patterns—and the ultra-low computational overhead demanded by resource-constrained edge terminals. To address this specific constraint, StarNet-RiceSeg is proposed as a lightweight semantic segmentation network explicitly tailored for unmanned harvesters. Initially, the architecture incorporates the minimalist StarNet as its backbone. By leveraging the unique “Star Operation,” it implicitly maps features into a high-dimensional nonlinear space, thereby significantly augmenting feature discriminability while drastically curtailing computational overhead. Furthermore, to mitigate the misdetection issues stemming from the textural similarity between lodged and upright rice, the Rice Spatial Attention (RSA) module was designed. By intensifying feature interaction within the spatial dimension, this module steers the network to focus on the cohesive morphology of lodged regions while effectively suppressing background noise. Experiments conducted on a self-constructed high-resolution rice lodging dataset demonstrate that StarNet-RiceSeg achieves a mIoU of 94.42%, significantly outperforming mainstream models such as U-Net, DeepLabV3+, SegNet and HRNet. Notably, the model maintains a compact footprint with only 8.01 million parameters and a computational load as low as 9.32 GFLOPs. Following optimization with TensorRT, the system achieved a real-time inference speed of 32.51 FPS on the NVIDIA Jetson Xavier NX embedded platform. These results indicate that StarNet-RiceSeg provides a high-precision, low-latency solution for perceiving rice lodging areas in complex field environments, facilitating unmanned precision harvesting.

Image feature detection serves as the cornerstone for numerous vision applications, and it has found extensive use in agricultural harvesting. Nevertheless, determining the optimal feature extraction technique for a specific situation proves challenging, as the Ground Truth correlation between images is exceedingly elusive in harsh agricultural harvesting environments. In this study, we assemble and make publicly available the inaugural agricultural harvesting dataset, encompassing four crops: rice, corn and soybean, wheat, and rape. We develop an innovative Ground Truth-independent feature detector assessment approach that amalgamates efficiency, repeatability, and feature distribution. We examine eight distinct feature detectors and conduct a thorough evaluation using the amassed dataset. The empirical findings indicate that the FAST detector and ASLFeat yield the most exceptional performance in agricultural harvesting contexts. This evaluation establishes a trustworthy bedrock for the astute identification and application of feature extraction techniques in diverse crop reaping situations.