Search Results (186)

The growth of Intelligent Transportation Systems (ITS) has made high-quality perception data from multi-sensor setups essential. Pairing millimeter-wave (mmW) radar with a monocular camera is a common way to recover three-dimensional information about the environment, but aligning the two is difficult because sparse radar point clouds and dense camera images differ sharply in how they sense a scene. The problem grows more severe in roadside infrastructure, where the high mounting elevation introduces perspective distortion that vehicle-mounted systems rarely face. This paper presents a systematic review, conducted under the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, of radar-camera extrinsic calibration for fixed roadside infrastructure, organizing existing work into a taxonomy that separates traditional two-stage pipelines from recent end-to-end learning frameworks. Because methods designed specifically for roadside units remain scarce, the review also covers vehicle- and robot-mounted methods whose static-sensor formulation carries over to fixed roadside deployment. For the two-stage pipeline, the analysis covers target-based and targetless correspondence registration along with the optimization techniques and algorithmic assumptions behind parameter estimation. The end-to-end learning literature shows a clear shift toward self-supervised and fusion-based models, some of which report real-time performance. The review also compares the metrics and procedures used to quantify calibration accuracy. Progress is evident, but robustness in cluttered urban environments remains an open challenge, and the paper closes by outlining future directions, arguing that standardized roadside benchmarks are needed before scalable, targetless calibration can mature. Full article

Modern fulfillment centers share floor space with human workers, making warehouse multi-robot task allocation a safety-critical problem. We propose SHARP (Safe Heterogeneous Allocation with Risk Prediction), a Transformer-based constrained reinforcement-learning framework with a closed-form deployment-time safety mask. Under a Gaussian pedestrian belief and fixed closest-approach directions, the mask uses Bonferroni-allocated per-pair CVaR scores; a nonnegative mask score implies a conservative trajectory-level chance constraint under the stated assumptions. We also present an idealized primal–dual surrogate analysis, without claiming global convergence for the nonconvex Transformer/PPO implementation. Expanded experiments use ten training seeds per learned method and deterministic final-checkpoint evaluation on twenty independently generated held-out instances. No statistically significant difference between SHARP and Lagrangian-PPO was detected in any of the four scenarios. The held-out analysis further reveals late-training instability and severe over-conservatism in the dense S40_high scenario. These findings position SHARP as an auditable geometric filtering mechanism, while identifying conservatism and training stability as important limitations for deployment. Full article

Driven by artificial intelligence, multi-source sensing, agricultural robots and big data technologies, global agriculture is rapidly upgrading from precision agriculture and agriculture 4.0 to agriculture 5.0. Artificial intelligence has evolved from a single diagnostic tool to an intelligent system that integrates the “perception-decision-execution” process throughout. It is widely applied in crop phenotype analysis, remote sensing monitoring, yield prediction, and autonomous operation of intelligent equipment, etc. This article takes the framework of “intelligent perception-cognitive decision-autonomous execution” to systematically review the core technologies, typical applications, and frontier directions of agricultural artificial intelligence. It focuses on introducing the progress of key technologies such as three-dimensional phenotype, hyperspectral remote sensing, multimodal fusion, and causal machine learning, as well as their value in improving resource utilization efficiency, enhancing climate resilience, and supporting field precision management. At the same time, it points out that current agricultural AI still faces practical bottlenecks such as insufficient generalization ability of models, scarce data and high annotation costs, difficulties in edge deployment, barriers in multi-source data integration, and weak interpretability and engineering reliability. Future research will focus on the construction of closed-loop autonomous farms, the collaboration of agricultural large models and intelligent agents, the construction of data centers and AI and data infrastructure, and the development of green and low-cost AI research. This will provide support for the technological innovation and industrialization implementation of agricultural artificial intelligence. Full article

This review provides a comprehensive synthesis of robotic fruit harvesting systems, with a particular focus on the system-level integration of perception, manipulation, and fruit detachment within autonomous harvesting environments. Recent advances in machine vision, deep learning, sensor fusion, robotic end-effectors, grasping strategies, and motion planning are critically analyzed alongside cutting, pulling, and vibration-based detachment mechanisms under unstructured orchard conditions. Beyond component-level analysis, this review emphasizes the critical role of perception–action coupling and highlights key system integration challenges, including localization errors, perception-to-action latency, and environmental variability, which continue to limit reliable field deployment. In addition, orchard and pre-harvest-related factors such as canopy structure, fruit distribution, and detachment force variability are examined in relation to their direct impact on system performance, robustness, and harvesting efficiency. Furthermore, the review extends toward system-level considerations by incorporating performance evaluation metrics, economic feasibility, and scalability constraints, which are essential for transitioning robotic harvesting systems from experimental prototypes to commercially viable solutions, including practical field deployment in distributed and multi-robot harvesting systems. Emerging technologies, including artificial intelligence, advanced sensing, digital agriculture, and energy-aware system design, are discussed as key enablers for achieving adaptive, data-driven, and scalable autonomous harvesting. The novelty of this work lies in proposing an integrated framework that explicitly links perception, manipulation, and detachment with orchard-level constraints and deployment requirements, thereby bridging the gap between algorithmic advancements and real-world implementation of autonomous fruit harvesting systems. Full article

The growing demand for elderly and bedridden patient care in hospitals, nursing homes, and long-term care facilities has increased the need for safe and efficient repositioning methods. Repositioning immobile patients is essential for preventing pressure injuries and other complications associated with prolonged immobility. However, this task is still commonly performed manually using bed sheets, pillows, and similar support aids, making it physically demanding and increasing the risk of musculoskeletal injury among caregivers. This paper presents a two-stage soft growing robot for safe and assisted patient repositioning from a supine posture to a side-lying position. The proposed mechanism consists of two soft pneumatic chambers with distinct roles. The first chamber enables pressure-driven eversion, allowing the robot to deploy smoothly beneath the patient with minimal friction. The second chamber is then pressurized to generate the lifting and rolling motion required for repositioning. A first-principles dynamic model of the pressure-driven vine robot is developed by integrating pneumatic supply dynamics, internal pressure evolution, and tip-extension mechanics within a Lagrangian framework. Based on this model, a robust multi-stage nonlinear model predictive control strategy is formulated to regulate deployment beneath the patient under parameter uncertainty. The rolling dynamics of the second stage are also analyzed to determine the minimum pressure required for repositioning as a function of patient weight and roll angle. Simulation results show that the proposed controller achieves smooth and accurate deployment while satisfying input and state constraints under uncertainty. The rolling analysis further indicates that the required pressure increases with patient weight and decreases with roll angle. These findings demonstrate the potential of the proposed mechanism to reduce caregiver effort and enable safe, controlled patient repositioning. Full article

Recent Vision–Language–Action (VLA) models have rapidly emerged as general-purpose robotic policies that integrate language understanding, visual perception, and robot control. However, prior studies and surveys have primarily emphasized backbone architectures, action decoders, training recipes, and benchmark performance, whereas relatively limited systematic attention has been given to sensor modality selection, heterogeneous signal alignment and fusion, and their connection to action generation, all of which are critical to the performance and safety of real-world robotic manipulation. This survey addresses this gap by reinterpreting VLA within the framework of a sensor–fusion–action pipeline. This study first presents a systematic taxonomy of major sensor modalities, including RGB, depth, tactile sensing, force/torque, proprioception and inertial measurement unit, multi-spectral/thermal, and event-based vision, and compares them in terms of the physical information they provide, their characteristic failure modes, and their deployment constraints. This survey further reviews teleoperation-, human video-, and simulation-based data collection pipelines, together with representative dataset configurations, and analyzes the multi-modal design space from a sensor-centric perspective, including early and late fusion, cross-attention, token-level fusion, adapters, mixture of experts, and multi-rate action representations. In addition, this study identifies a strong bias in existing benchmarks toward RGB-centric inputs and single success-rate metrics and emphasizes the need for a multidimensional evaluation framework incorporating robustness, worst-case performance, safety, latency, and efficiency. By shifting the focus away from a model-centric narrative and explicitly accounting for real-world sensor complexity, this survey seeks to establish a sensor-centered foundation for the next generation of Physical AI. Full article

Tomato leaf diseases substantially reduce tomato yields and quality and remain a persistent challenge for efficient crop management. Although deep learning-based detectors have achieved strong accuracy in controlled benchmarks, many existing solutions are still difficult to transfer to resource-constrained agricultural systems because they rely on high-end GPUs, consume considerable power, and often lose performance after deployment on embedded devices. To address this practical gap, this study proposes HGS-YOLO, a system-oriented deployable lightweight adaptation of YOLOv11 for leaf-level tomato disease detection, together with an end-to-end edge sensing pipeline for low-power agricultural deployment. The main contribution lies in the coordinated system-level co-design of model structure, optimization, and deployment rather than in a novel detector architecture. Specifically, YOLOv11 is adapted through three coordinated modifications: an HGNetV2 backbone for efficient feature extraction, an HS-FPN neck with channel attention for lightweight multi-scale fusion, and an MPDIoU loss function for more stable localization optimization. Beyond the model architecture, the study establishes a complete engineering pipeline that includes training, optimization, post-training quantization, and hardware deployment with BPU acceleration on a D-Robotics RDK X5 handheld platform. Comprehensive benchmark experiments indicate that HGS-YOLO achieves 93.6% ${\mathrm{mAP}}_{50}$ and 72.1% mAP@[0.5:0.95] with 86.5% recall, only 1.3 M parameters, and a 3.1 MB model size, substantially reducing the model complexity and storage cost relative to the YOLOv11 baseline. A three-seed retraining comparison shows that HGS-YOLO trades roughly 0.5 ${\mathrm{mAP}}_{50}$ points for this compactness (a statistically significant but small concession) and recovers the cost on the deployment side: on the RDK X5 chip, HGS-YOLO is the fastest, most memory-efficient, and lowest-power model among all compared detectors. Indoor deployment tests using separately collected tomato leaf samples further achieve 90.3% ${\mathrm{mAP}}_{50}$, 82.3% recall, 89.0% precision, 25.0 ± 0.4 ms end-to-end latency, 40.0 ± 0.6 FPS, and 9.8 ± 0.4 W average system power. After PTQ, the ${\mathrm{mAP}}_{50}$ drops from 93.6% to 93.0% on the same benchmark; because this figure was measured under controlled imaging conditions, it is presented as an in-distribution reference point rather than as evidence of robustness in the open field. We also took the handheld system into a working tomato greenhouse for a small outdoor field round, where it ran end-to-end and produced on-device disease detections under natural sunlight, specular highlights, partial occlusion, background clutter, and handheld motion blur. These results show that HGS-YOLO reaches a good balance of accuracy, efficiency, and deployability and that it works in the field on an independent small-scale test; validating it more widely across sites, seasons, and weather is left to future work. Full article

In response to the challenges of large fruit-scale variation, dense target distribution, severe leaf occlusion, and complex backgrounds in green plum detection within orchards, this paper proposes a lightweight multi-scale edge geometry-aware network (MEGNet). First, the Green Plum Detection Dataset (GPD) is constructed to provide realistic orchard scene data for the task. Next, we enhance the model’s structure based on YOLO11n by designing an efficient multi-scale feature fusion attention module (EMFFA) to improve the expression of multi-scale fruit features. We also introduce a color-edge guided dual-discriminator feature enhancement module (CED) to strengthen feature discrimination in complex backgrounds. A coordinate attention ghost detection head (CAGDetect) is proposed to reduce model parameters and computational complexity. Additionally, a geometry-consistency modulated CIoU loss function (GC-CIoU) is introduced to improve target localization stability in occluded and dense scenes by incorporating a geometric consistency modulation mechanism. Experimental results show that on the GPD, MEGNet achieves a Precision of 93.9%, Recall of 86.2%, mAP₅₀ of 93.2%, and mAP_50:95 of 76.1%. The model’s Parameters are only 2.13 M, with FLOPs of 4.7 G. Compared to the baseline YOLO11n model, Precision, Recall, mAP₅₀, and mAP_50:95 are improved by 2.5%, 5.2%, 4.4%, and 4.6%, respectively. Additionally, deployment experiments on the Jetson Orin Nano embedded device demonstrate real-time detection speeds of 31–33 FPS. The proposed method provides an efficient and reliable solution for intelligent harvesting systems, orchard monitoring platforms, and agricultural robot vision perception. Full article

Embedding large language model (LLM) coordinators in production electronic systems, connected vehicles, multi-robot fabrics, IoT control loops, telecommunications orchestration, demands a pre-delivery filter stage that preserves ethical guarantees under adversarial influence at deployment scale. We present a constitutional governance layer that filters compiled influence policies before they reach a heterogeneous population of grounded LLM agents whose hybrid decision model combines a game-theoretic base probability with an LLM-evaluated narrative shift attenuated by per-agent resistance. Four experiments on a Barabási–Albert scale-free network of 30 agents powered by Llama-3.3-70B-Instruct show that the filter holds an Ethical Cooperation Score (ECS) of $0.176$ (multi-seed mean $0.163$, 95% confidence interval (CI) $[0.150,0.174]$) against an unconstrained baseline of $\mathrm{ECS}=0$, enforced by a hard integrity gate ($1.000$ vs. $0.000$). We surface an autonomy paradox in which unconstrained agents resist manipulation more forcefully ($0.856$ vs. $0.728$) yet collapse to $\mathrm{ECS}=0$, establishing that system-level integrity cannot be delegated to agent-level defence. The advantage is monotonic in resistance ($+0.174$ to $+0.183$), seed-stable (Cliff’s $\delta =1.0$, complete separation), topology- and backbone-invariant across five contemporary LLMs, robust to alternative ECS formulations, and reproduces at N = 100. Against constitutional artificial intelligence (CAI) critique-revise and LlamaGuard-style safety-classifier baselines, the framework matches the integrity floor and adds a measurable margin on the secondary risk surface (burst timing, composite manipulation risk). The filter runs at 0.78 μs/call ($\approx 1.3\times {10}^6$ decisions/s/core), supporting always-on deployment as a stateless, model-agnostic component of LLM agent pipelines in adversarially contested electronic systems. Full article

The field inspection and ripeness detection of tomatoes in China remain heavily dependent on manual labor, while existing robotic solutions often exhibit limited functionality, poor environmental adaptability, prohibitive hardware costs, and unstable positioning accuracy. To address these limitations, this study proposes an intelligent tomato inspection robot that seamlessly integrates real-time ripeness recognition with precise spatial localization. Built upon a Raspberry Pi 5 core controller, the robot employs a lightweight, layered modular architecture designed to flexibly navigate complex agricultural environments. A comprehensive, multi-dimensional image dataset of tomato ripeness was constructed to train a three-category detection model based on the YOLOv8n architecture. Following 413 training epochs, the model demonstrated exceptional performance, achieving an overall mAP@0.5 of 87.8% and an mAP@0.5:0.95 of 72.7% on the held-out test dataset. In field inspections, the system achieved detection precisions of 82.22% for immature tomatoes, 92.66% for half-ripened tomatoes, and 100% for fully ripe tomatoes, successfully identifying all ripe tomatoes and satisfying the practical demands of field inspection. Furthermore, the integration of an Ultra-Wideband positioning system yielded an overall Root Mean Square Error of 0.231 m, successfully confining positioning errors to within 0.24 m to fully satisfy the stringent localization demands of crop-level inspection. Field evaluations confirmed that under optimal configurations, the robot can efficiently inspect a 50-m planting row in 10 min (±1 min) and maintains a continuous operational battery life of 2 h (±10 min). The core contribution of this work is the system-level integration and optimization of technologies for greenhouse agriculture. This integrated design achieves low hardware cost and high deployment flexibility, addressing longstanding challenges of labor-intensive inspection and delayed harvesting, and delivering a practical solution for intelligent tomato plantation management. Full article

With the continuous growth of global population and increasing pressure on food security, the transformation toward precise and intelligent agricultural production has become an inevitable trend. In this context, accurate identification of field weeds is crucial for improving crop yields and reducing agricultural inputs. However, agricultural Internet of Things (IoT) edge devices are generally subject to strict constraints in terms of power consumption, storage, and real-time performance. Existing lightweight convolutional neural networks often struggle to simultaneously achieve high accuracy and low resource consumption for fine-grained weed identification tasks. To address this challenge, this paper proposes a hardware aware lightweight convolutional neural network named Weed-Aware Ghost Enhanced Network (WAGENet) for microcontroller deployment. The network synergistically integrates Ghost low-cost feature generation, Mobile Inverted Bottleneck Convolution (MBConv) for deep semantic extraction, Squeeze and Excitation (SE) and Coordinate Attention (CA) dual attention mechanisms for channel space joint calibration, and Atrous Spatial Pyramid Pooling (ASPP) for multi-scale context fusion. It constructs a progressive feature abstraction system from shallow textures to high-level semantics. On the public DeepWeeds dataset, WAGENet achieves 95.71% classification accuracy and 93.80% F1 score with only 0.163 M parameters and 2.43 × 10⁸ multiply accumulate operations (MACC), attaining a parameter efficiency of 587.19%/M and significantly outperforming existing mainstream lightweight models. The model has been successfully deployed on the STM32H7B3I microcontroller development board, achieving a single inference latency of 94.63 ms, an internal Flash footprint of only 686.95 KiB, and a single inference energy consumption of 41.45 mJ. Experimental results demonstrate that WAGENet achieves a trade off among accuracy, latency, and energy consumption under strict resource constraints, providing a reproducible microcontroller deployment paradigm for battery powered field robots, drones, and other agricultural IoT edge devices. Full article

Underwater image enhancement is critical for marine robotic perception, yet existing methods often face a persistent trade-off between restoration quality, structural reliability, and deployment efficiency. Although lightweight enhancement networks are attractive for resource-constrained underwater platforms, many of them mainly rely on empirical architectural simplification and appearance-oriented objectives, with limited mathematical analysis of complexity reduction, semantic regularization, and optimization coordination. To address this issue, this paper proposes MobileGAN, a lightweight underwater image enhancement framework equipped with dual-reference regularization and a theoretical analysis module. The proposed generator adopts a compact encoder–bottleneck–decoder architecture based on depthwise separable convolutions, which substantially reduces convolutional redundancy while preserving effective restoration capability. A dual-reference feature consistency formulation is introduced to simultaneously constrain the enhanced image toward the high-quality target representation and the degraded-input semantic anchor. In addition, an edge-aware regularization term and a stage-wise dynamic weighting mechanism are incorporated to improve local structure recovery and multi-objective optimization behavior. Beyond architectural design, we provide a mathematical analysis of the proposed framework from three aspects: computational complexity reduction, geometric interpretation of dual-reference regularization, and piecewise optimization properties of stage-wise weighted training. Extensive experiments on the UIEB benchmark demonstrate that MobileGAN achieves a favorable trade-off between enhancement quality and computational efficiency. The proposed method maintains real-time inference with a compact model size while providing competitive structural consistency and detail restoration. These results indicate that MobileGAN is not only a practical deployment-oriented enhancement framework but also an interpretable optimization model with analyzable structural properties. Full article

Robotics integration across manufacturing, healthcare, and hazardous environments demands robust performance evaluation. This study proposes a comprehensive Task–Environment–System–Metric (TESM) framework to link operational tasks and environmental constraints with quantifiable metrics. Based on TESM, a multi-level evaluation system is established, covering kinematic/dynamic performance, perception, human–robot interaction (HRI), reliability, and lifecycle economics. We systematically review key evaluation methodologies, including mechanistic modeling, digital twin simulation, physical testing, and multi-criteria decision-making (MCDM). Furthermore, typical engineering applications—ranging from industrial manipulators and mobile robots to collaborative and field systems are analyzed to demonstrate practical implementation. Despite significant progress, challenges persist regarding unified standards, testing fidelity, and the “black box” nature of data-driven assessments in safety-critical scenarios. This review concludes by identifying future research directions, such as establishing benchmark testing platforms, improving lifecycle assessment schemes, and developing modular evaluation tools. These advancements aim to ensure the scalable and reliable deployment of robotic systems in complex engineering environments. Full article

Robot-assisted telerehabilitation (RAT) combines rehabilitation robotics with digital health workflows to extend access to upper-limb (UL) therapy after stroke. Mixed reality (MR) may support therapist–patient interaction and task visualization; however, early-stage systems require rigorous evaluation of safety and usability before deployment in the home. In a formative, mixed-methods usability study conducted in a controlled setting using a telerehabilitation workflow, six individuals post-stroke (≥3 months) and six occupational therapists (OTs) completed a single supervised session with a desktop-mounted end-effector type therapeutic robot (iTbot) integrated with Microsoft HoloLens 2. Participants performed structured passive and active UL exercises while therapists supervised and interacted with the system via the MR control interfaces. Safety was evaluated by documenting observed adverse events and safety-stop activations. Usability and user experience were assessed using the System Usability Scale (SUS), study-specific satisfaction questionnaires (reported with scale ranges), and semi-structured follow-up interviews analyzed using thematic analysis. All participants completed the session without observed adverse events or safety-stop activations. Overall usability was favorable, with a mean (SD) SUS total score of 78.3 (15.9) out of 100 (stroke: 74.2 [18.1]; occupational therapists: 82.5 [13.5]). Qualitative feedback indicated that MR was perceived as engaging and intuitive by many users, while also identifying implementation needs relevant to real-world telerehabilitation, including clearer onboarding, simplification of certain MR interactions, and improved physical interfaces (e.g., handle options). Therapists highlighted workflow considerations for remote supervision and patient independence. Together, these findings support progression to multi-session, in-home studies to quantify remote assistance needs, technical reliability, adherence, and clinical outcomes. Full article

To address the dual challenges of discriminating weeds from rice seedlings for precision weed management operations, such as targeted spraying and robotic weeding, in complex paddy-field scenes and deploying high-precision models on resource-limited edge devices, we propose DGS-Net, a deformable attention, GSConv-based feature fusion, and SEAM-enhanced lightweight network based on YOLOv11n. The backbone incorporates a convolutional block with parallel split attention and deformable attention transformer (C2PSA_DAT) module to improve the extraction of irregular and fine-grained weed features, the neck integrates a VoV-GSCSP module to enable lightweight multi-scale feature fusion for small and densely distributed targets, and a separated and enhancement attention module (SEAM) is placed before the detection head to enhance robustness under leaf occlusion and complex paddy-field background interference. In comparative experiments conducted on the paddy-field dataset under unified training and evaluation settings, DGS-Net achieved 91.7% precision, 86.8% recall, and 92.4% mean average precision (mAP), with a model size of 5.8 MB and a computational cost of 6.2 giga floating-point operations (GFLOPs). Compared with representative lightweight baseline detectors, DGS-Net showed a more favorable balance between detection accuracy and deployment efficiency. In additional edge-device deployment tests using the test set, the model sustained real-time inference at 32.5 FPS and achieved mAP@0.5, precision, and recall of approximately 0.928, 0.919, and 0.867, respectively. Overall, DGS-Net improves irregular feature extraction, enables lightweight multi-scale feature fusion, and increases robustness to occlusion while retaining strong deployability. The method therefore provides practical visual-perception support for precise, real-time crop–weed discrimination and precision weed management in complex paddy-field environments. Full article