Wuhu Hit Robot Technology Research Institute

Fracture injuries often lead to complex bone fragmentations, posing significant challenges for accurate segmentation in surgical planning and trauma assessment. Manual annotation of each fragment is time-consuming and inconsistent, while existing automated methods often fail to separate individual fragments due to the wide variation in fracture types, irregular fracture surface, and close inter-fragment contact. To address these challenges, we introduce FracSegmentator, a deep learning approach for bone fragment instance segmentation. The model takes extracted bone regions in CT as input and isolates individual fragments by identifying fracture surfaces and separating closely contacting structures. Central to our approach is a Trauma-Prior-Guided Contrastive Learning module, which incorporates clinical knowledge through memory-based attention to better distinguish fractured surfaces from healthy regions. We evaluate FracSegmentator on four datasets that cover a range of anatomical sites and fracture patterns. The method achieves state-of-the-art results across all datasets and demonstrates strong generalization capabilities. By delivering accurate and efficient fragment-level segmentation, FracSegmentator supports critical downstream tasks such as automated fracture diagnosis, surgical planning, and preoperative reduction simulation.

ABSTRACT Research on autonomous driving in unstructured outdoor environments is less advanced than in structured urban settings due to challenges like environmental diversities and scene complexity. These environments–such as rural areas and rugged terrains–pose unique obstacles that are not common in structured urban areas. Despite these difficulties, autonomous driving in unstructured outdoor environments is crucial for applications in agriculture, mining, and military operations. Our survey reviews over 250 papers for autonomous driving in unstructured outdoor environments, covering offline mapping, pose estimation, environmental perception, path planning, end‐to‐end autonomous driving, datasets, and relevant challenges. We also discuss emerging trends and future research directions. This review aims to consolidate knowledge and encourage further research for autonomous driving in unstructured environments. To support ongoing work, we maintain an active repository of up‐to‐date literature and open‐source projects at: https://github.com/chaytonmin/Survey‐Autonomous‐Driving‐in‐Unstructured‐Environments .

Plant phenotyping is essential for elucidating genotype–environment interactions, yet conventional methods remain labor-intensive and low-throughput. TraitDiscover transcends these constraints by uniting multimodal sensing with tightly coupled hardware-software orchestration in a single, end-to-end phenotyping platform. Aligned with the ”Plant Phenotyping Trinity” framework, the system comprises a millimetre-accurate triaxial automation unit, a modular sensor array–RGB imaging, three-dimension laser scanner or LiDAR (3D), infrad (IR) thermal imaging, hyperspectral imaging (HSI), and photosynthesis (PS) imaging–and the dedicated software TraitNavigator suite into one cohesive system. A unified spatiotemporal synchronization mechanism enables robust time-series analysis and fusion of multisource phenotypic data across the entire crop growth period, while the DepthCropSeg algorithm and a night-time imaging module enhance trait extraction under complex conditions, providing G × E × P-ready, multimodal phenotypic datasets. Validation across soybean, maize, and rice trials demonstrated high sensitivity—detecting drought stress four days before visible symptoms, identifying glyphosate injury 24 hours ahead of manual scoring, and quantifying local adaption patterns across ecological gradients. While challenges remain in scaling to complex open-field conditions, TraitDiscover offers a scalable, data-driven approach to accelerate stress phenotyping and breeding decisions and is readily poised for deeper integration with AI to advance sustainable agriculture.

Abstract This paper revisits the group controllability of two‐time scales multi‐agent systems, aiming to achieve innovation from the perspective of heterogeneity and switching topology. To accomplish this objective, the singular perturbation system theory and boundary layer theory are introduced to model the systems. Then, Wonham's Geometric Methods alongside concepts, such as invariant subspaces and controllable state sets, are employed to establish criteria for group controllability. Furthermore, we propose a method to construct switching sequences aimed at ensuring group controllability. Numerical simulations validate the effectiveness of the derived conclusions.

The widespread deployment of automated guided vehicles (AGVs) in dynamic complex environments poses a critical challenge in achieving efficient avoidance and energy optimization. This paper proposes an observation-enhanced reinforcement learning with reward constraints algorithm (ORLRCA) to enhance AGV safety and energy efficiency obstacle avoiance in complex scenarios. In our work, firstly, a multimodal perceptual attention mechanism is introduced to dynamically capture obstacle motion patterns and environmental semantic features, thereby enhancing scene perception capabilities. Secondly, a multi-scale prior reward-constrained function is designed to jointly optimize safety distance, path smoothness, and energy consumption metrics, effectively addressing the suboptimal convergence caused by conflicting objectives in RL strategies. Finally, leveraging the actor-critic network architecture of proximal policy optimization (PPO), we achieve end-to-end optimization of robust obstacle avoidance policy generation through synergistic integration of attention-enhanced state representations and multi-scale reward signals. To validate efficacy, a high-fidelity simulation environment is developed for comparative experiments. Results demonstrate that the proposed algorithm exhibits superior performance in obstacle avoidance success rate and energy efficiency compared to baseline RL methods, establishing a theoretical foundation for secure deployment and energy-efficient management of AGVs in practical industrial scenarios.

Birds' extraordinary aerial agility and environmental interaction enable complex tasks such as mid-air hunting, perching, and nest-building, inspiring the development of advanced aerial robots with similar manipulation capabilities. However, existing platforms often face challenges such as large size, heavy payloads, end-effector torque interference, and limited functionality, severely restricting their practical deployment. Drawing inspiration from the biological, structural, and actuation characteristics of human hands, we propose a hand-like robot that integrates flight and grasping, demonstrating the synergistic advantages of compact structure, agile flight, and versatile manipulation. We propose an autonomous framework including efficient mission planning and multi-level adaptive control, enabling the robot to precisely and smoothly perform human-like grasping, opening doors, forest perching, object transport, and interactive tasks. Additionally, the framework supports human-robot collaboration, empowering individuals with mobility impairment to conduct remote transportation and airborne operations. Outdoor tests, which include perching in various scenarios, navigating confined spaces, and transporting payloads across challenging terrain, validate the proposed vehicle's potential in aerial delivery and manipulation tasks. These results demonstrate emerging possibilities for aerial operation, assistance, and delivery with integrated flight and manipulation abilities.

Accurately characterizing the structural state of membrane reflector antennas (MRA) remains challenging due to the difficulty in determining stress distribution through geometric measurement alone. Although photogrammetry provides high-precision geometric data, it falls short of capturing mechanical pre-tension and is notably influenced by gravity, which limits its utility in guiding surface accuracy adjustments. This paper proposed an integrated approach combining photogrammetry with a nonlinear finite element method (NFEM) to achieve high-fidelity imaging and effective shape adjustment of electrostatically formed MRA, explicitly accounting for gravity effects during ground-based measurement and shape control. The proposed method establishes a mechanical model that incorporates real-world geometric data under gravity and performs force–shape matching to reconcile discrepancies between physical and simulation models. Experimental validation demonstrates that the gravity-corrected NFEM model closely aligns with the physical antenna, with a deviation in surface accuracy within 9.9%. Using this refined model, we successfully optimized electrode voltages and cable tensions, improving the surface accuracy of the physical model from an initial 0.7033 mm to 0.5723 mm. This work provides a reliable and efficient strategy for the shape control and adjustment of membrane space structures under gravity, with potential applications in large deployable antennas, solar sails, and other tension-controlled flexible systems.

Humans could sensitively perceive and identify objects through dense mechanoreceptors distributed on the skin of curved fingertips. Inspired by this biological structure, this study presents a general conformal integration method for flexible tactile sensors on curved fingertip surfaces. By adopting a spherical partition design and an inverse mode auxiliary layering process, it ensures the uniform distribution of stress at different curvatures. The sensor adopts a 3 × 3 tactile array configuration, replicating the 3D curved surface distribution of human mechanoreceptors. By analyzing multi-point outputs, the sensor reconstructs contact pressure gradients and infers the softness or stiffness of touched objects, thereby realizing both structural and functional bionics. These sensors exhibit excellent linearity within 0-100 kPa (sensitivity ≈ 36.86 kPa-1), fast response (2 ms), and outstanding durability (signal decay of only 1.94% after 30,000 cycles). It is worth noting that this conformal tactile fingertip integration method not only exhibits uniform responses at each unit, but also has the preload-free advantage, and then performs well in pulse detection and hardness discrimination. This work provides a novel bioinspired pathway for conformal integration of tactile sensors, enabling artificial skins and robotic fingertips with human-like tactile perception.