Within the domain of intelligent picking robotics, fruit recognition and positioning are essential. Challenging conditions such as varying light, occlusion, and limited edge-computing power compromise fruit maturity detection. To tackle these issues, this paper proposes a lightweight algorithm YOLO-ELS based on YOLOv8n. Specifically, we reconstruct the backbone by replacing the bottlenecks in the C2f structure with Edge-Information-Enhanced Modules (EIEM) to prioritize morphological cues and filter background redundancy. Furthermore, a Large Separable Kernel Attention (LSKA) mechanism is integrated into the SPPF layer to expand the effective receptive field for multi-scale targets. To mitigate occlusion-induced errors, a Spatially Enhanced Attention Module (SEAM) is incorporated into the decoupled detection head to enhance feature responses in obscured regions. Finally, the Inner-GIoU loss is adopted to refine bounding box regression and accelerate convergence. Experimental results demonstrate that compared to the YOLOv8n baseline, the proposed YOLO-ELS achieves a 14.8% reduction in GFLOPs and a 2.3% decrease in parameters, while attaining a precision, recall, and mAP@50% of 92.7%, 83.9%, and 92.0%, respectively. When compared with mainstream models such as DETR, Faster-RCNN, SSD, TOOD, YOLOv5s, and YOLO11n, the mAP@50% is improved by 7.0%, 4.7%, 11.4%, 8.6%, 3.1%, and 3.2%. Deployment tests on the NVIDIA Jetson Orin Nano Super edge platform yield an inference latency of 25.2 ms and a detection speed of 28.2 FPS, successfully meeting the real-time operational requirements of automated harvesting systems. These findings confirm that YOLO-ELS effectively balances high detection accuracy with lightweight architecture, providing a robust technical foundation for intelligent fruit picking in resource-constrained greenhouse environments.

This paper addresses fault estimation in discrete-time semi-Markov jump neural networks (s-MJNNs) under the Round-Robin protocol and proposes an innovative extended state observer-based approach. Unlike studies considering only constant transition rates, this work investigates s-MJNNs with time-varying transition probabilities, which more closely reflect practical situations. By incorporating actuator and sensor faults as augmented state variables, an extended state observer is proposed to estimate system states and faults simultaneously. To alleviate network congestion and optimize communication resources, the Round-Robin protocol is adopted to schedule data transmission efficiently. By constructing a Lyapunov–Krasovskii functional and applying the discrete Wirtinger inequality, sufficient conditions are derived to ensure the mean square exponential stability and dissipative performance of the system. The observer gain parameters are computed using the linear matrix inequality (LMI) method. Numerical simulations validate the effectiveness and performance of the proposed fault estimation method.

Selective harvesting of greenhouse cucumbers requires accurate detection with low inference latency. In greenhouse canopies, mature cucumbers are often partly occluded and visually similar to surrounding stems and leaves, which makes harvestability recognition difficult. Existing real-time detectors still struggle to preserve fine boundary cues, capture long-range context, and remain compact enough for on-device inference under these conditions. This study proposes MCS-DETR, an efficient multi-scale context-aware detector built on RT-DETR. Instead of increasing model scale, MCS-DETR redesigns shallow feature extraction, high-level contextual interaction, and cross-scale feature aggregation within a compact framework. A shallow feature level is also retained to preserve fine contour information. On the greenhouse cucumber dataset, MCS-DETR achieved 93.4% mAP@0.5 and 76.8% mAP@0.5:0.95, outperforming RT-DETR-R18 while requiring fewer parameters and less computation. On an NVIDIA Jetson Orin NX Super (Hunan ChuangLebo Intelligent Technology Co., Ltd., Room 2003, Building C, Xinchanghai Digital Center, Changsha Economic and Technological Development Zone, Changsha, Hunan, China) platform, it reached 26.3 FPS after TensorRT acceleration. These results indicate that MCS-DETR can provide an efficient on-device perception module for real-time greenhouse cucumber detection.

In this paper, the fixed-time fault-disturbance estimation problem for an emergency rescue quadrotor UAV during the takeoff–mission execution–landing process is investigated. Considering the complex operating conditions of rescue missions, actuator efficiency loss, actuator bias faults, rescue-environment airflow disturbances, payload-induced disturbances, near-ground aerodynamic effects, and unmodeled dynamic uncertainties are uniformly represented as a lumped fault disturbance entering the angular-velocity dynamic channel. A multi-stage disturbance model is first established according to the takeoff stage, mission execution stage, and landing stage. Then, a fixed-time fault-disturbance estimator with stage-dependent gains is designed to estimate the lumped disturbance. Based on matrix Lyapunov functions and fixed-time stability theory, it is proven that the angular-velocity estimation error and the lumped fault-disturbance estimation error can enter and remain in a small bounded neighborhood within a fixed time independent of the initial conditions. Moreover, the bounded disturbance jumps at the stage-transition instants are analyzed, and the post-transition recovery property of the estimator is established. The simulation results under a practical emergency rescue scenario show that the proposed estimator can reconstruct the main variation trends of the roll, pitch, and yaw channel lumped disturbances, keep the estimation error bounded, and recover after stage-transition-induced disturbance variations.

The actuator-fault estimation problem is addressed for discrete-time semi-Markov jump neural networks subject to time-varying delays, external disturbances, and communication constraints induced by the weighted try-once-discard (WTOD) protocol. Under this protocol, only the measurement channel with the largest weighted error is transmitted at each sampling instant, while the unselected channels retain their previously stored measurements at the filter side. To estimate the actuator fault, a fault-weighting dynamic system is first introduced. Then, by incorporating the WTOD-induced held measurement into the state vector, an augmented estimation model is constructed to describe the fault-weighting dynamics and the protocol-induced data-holding behavior within a unified framework. Based on this model, a mode-dependent and channel-dependent reduced-order fault-estimation filter is designed. The distinctive feature of the proposed framework is that the reconstruction of selected state components and the estimation of the actuator fault are addressed within a unified reduced-order estimator whose parameters depend jointly on the semi-Markov mode and the active WTOD transmission channel. By employing a Lyapunov–Krasovskii functional and using the semi-Markov transition information together with the WTOD scheduling constraint, sufficient LMI-based conditions are derived to ensure mean-square exponential stability and strict $(T_1,T_2,T_3)-\delta$ dissipativity performance of the resulting estimation error system. Finally, two examples are provided to illustrate the numerical effectiveness of the proposed actuator-fault estimation method under different semi-Markov switching realizations.