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.

Rice–wheat combine harvesters face coupled traction and material-processing loads whose variability complicates energy saving without reducing throughput or harvest quality. This review connects five perspectives: intelligent sensing, electrified actuation, hybrid powertrain architecture, field-condition prediction, and energy management. It examines interfaces under common quality and battery energy boundaries. Forward crop sensing, feed-rate monitoring, and component-load observation provide complementary information, while independent electric drives make drum, fan, and sieve settings more controllable. Series, parallel, and power-split architectures offer different compromises between controllability, conversion losses, packaging, and sustained-load performance. Direct evidence remains limited: one field prototype reported 19.5% lower fuel consumption than a conventional harvester, a study-specific result that does not establish a general saving range. Evidence from tractors and other vehicles supports transferable methods but cannot replace rice–wheat validation. A comparison matrix examines rule-based, instantaneous, optimal, predictive, and learning-based energy management. The framework couples fuel use and clean-grain throughput with loss, breakage, impurity, and state-of-charge constraints, and outlines life-cycle cost and carbon accounting. It is methodological rather than a new numerical assessment. The priorities are synchronized field duty cycles, uncertainty-aware coordination, thermal and fault validation, and comparisons across sites and seasons. The synthesis supports powertrain design and controller development for more energy-efficient and intelligent harvesting.