We report a compact, filter-free hyperspectral imaging system via hardware-algorithm co-design of a liquid crystal micro-lens array (LC-MLA) and a digital prism. By jointly optimizing the LC-MLA voltage sequence and dispersion kernel within a multi-objective framework, our method achieves adaptive alignment between optical encoding and spectral reconstruction without mechanical scanning. Experiments demonstrate a peak signal-to-noise ratio of 34.39 dB and a spectral angle mapper of 1.463 degrees over 400-900 nm, enabling robust discrimination of metameric materials.

The evolution height and distribution pattern of mining-induced fractures beneath the hard roof of a deep ultra-thick coal seam are important parameters for the prevention and control of roof water hazards. This study took the 2303 working face of a coal mine as the engineering background. By integrating DEM numerical simulation, image processing technology, and field measurements, the typical failure structures of the mining-induced overburden and the evolution of the fracture field were quantitatively characterized from the perspectives of fracture count, fracture length, fracture area, fracture porosity, fracture density, fractal dimension, fracture entropy, fracture inclination angle, and permeability. On this basis, the mining-induced failure and movement characteristics of the overburden structure and the evolution of the fracture field beneath the hard roof of a deep ultra-thick coal seam were elucidated. The following beneficial findings were yielded: 1) The mining-induced overburden at different stratigraphic levels undergoes the transition from intact suspended structures to fractured and collapsed ones and from stable suspension to fracturing and rotational subsidence, and ultimately to bending subsidence. The water-conducting fracture zone was observed to sequentially go through five stages and reach a peak height of 224.0 m, with a fracture-to-mining ratio of 18.7. The simulation results are in good agreement with the field observations. 2) The evolution of fracture porosity and fractal characteristics associated with the upward propagation of fracture “opening, closure, and compaction” in different sub-regions of the mining-induced fracture field was revealed. The fractal dimension of the fracture field successively undergoes a dimension-rising stage, a dimension-stable stage, a dimension-falling stage, and another dimension-stable stage, and the fracture field displays a trapezoidal distribution along the mining direction. 3) The anisotropy of the spatial distribution of mining-induced fractures was quantified based on the fracture entropy index. When secondary fractures propagate along the main fracture path, the fracture entropy evolves nonlinearly. The compaction effect in the goaf causes partial fracture closure, leading to dynamic convergence of fracture entropy. In addition, the distribution and evolution characteristics of fracture inclination angle types in different regions and at different mining distances were statistically analyzed. 4) Considering the structural characteristics of the mining-induced overburden, a seepage model comprising a weak-seepage zone, an intermediate seepage transition zone, and a strong-seepage zone was constructed. Moreover, the evolution of mining-induced overburden seepage fractures was concluded, and the overburden damage and hazard-causing mechanism of ultra-thick coal seam mining was explained from both macroscopic and microscopic perspectives.

Under the condition of slicing mining in extra-thick coal seams, the presence of irregular coal pillars is likely to cause stress redistribution in the roadway region and induce rockburst. Taking the "3·22" rockburst event that occurred in the haulage roadway of the 250,101-2 working face in Huating Coal Mine as the engineering background, this paper investigated the occurrence mechanism of roadway rockburst under irregular coal pillar conditions by combining source mechanism inversion, numerical simulation, and theoretical analysis. The results show that the double-couple component is dominant in the moment tensor inversion results, indicating that the source type of this event was shear-type, and that the essence of the rockburst instability was the sudden shear slip of the coal-rock mass under high-stress conditions. The PFC simulation results show that, under the control of an L-shaped irregular coal pillar formed by a 20 m residual section pillar and a 34 m residual pillar, the overburden load developed a zonal load transfer pattern. Specifically, the 20 m pillar constituted the main load transfer channel, while the compacted zone above the 34 m pillar regained a certain bearing capacity after compaction of the caved rock mass and exerted an auxiliary reloading effect on the underlying surrounding rock, resulting in the haulage roadway not being in a fully destressed state. On this basis, a two-segment bearing model of the L-shaped irregular coal pillar was established, and the static stress distribution characteristics in the roadway region under the combined action of the two-segment loads were analyzed based on half-plane elasticity theory. Furthermore, by incorporating the attenuation law of vibration waves, a stress increment estimation model under dynamic loading disturbance was established, and the dynamic stress increment generated on the roadway surface by the "3·22" rockburst event was calculated to be about 4.06 MPa. Finally, the stress concentration characteristics and stress deflection effect under the control of the L-shaped coal pillar structure were discussed. The results show that an increase in the right-wing thickness of the L-shaped coal pillar structure enlarges the stress deflection zone and enhances stress redistribution toward the roadway, thereby increasing the possibility of rockburst under the combined action of high static stress and dynamic disturbance. The research results reveal the occurrence mechanism of roadway rockburst under L-shaped irregular coal pillar conditions, and can provide a theoretical reference for identifying rockburst hazard zones and optimizing working face layout parameters under similar engineering conditions.

The existence of discontinuous and time-varying singularities poses a fundamental challenge to phase modulation and imaging performance within optical systems. We first present a liquid crystal on-chip optical computing system (LCOC) integrating a liquid crystal microlens array (LC-MLA) and a polymer-dispersed liquid crystal (PDLC) film. The LCOC architecture comprises a cascade of two optical convolutional layers and a nonlinear activation layer, enabling programmable optical computation. By exploiting the electronic tunability of the LC-MLA for adaptive linear filtering and the scattering nonlinearity of the PDLC film, the system transforms discontinuous singularity distributions into continuous functions. Through optoelectronic co-optimization, our approach achieves real-time singularity modulation. Experimental results show that the LCOC suppresses singularities with a root-mean-square error (RMSE) of 0.03725, improving imaging fidelity. The system outperforms both single-layer optical and digital neural networks, offering a promising route toward high-performance, adaptive optical computing.

Event cameras offer microsecond-level temporal resolution and a wide dynamic range, but the sparse events generated in static scenes limit event-based high dynamic range (HDR) imaging. In this study, we propose a liquid crystal modulated event camera (LC-MEC) system, an integrated imaging system comprising a main lens, a liquid crystal microlens array (LC-MLA), and a camera for HDR reconstruction in static scenes. By applying discrete voltages to the LC-MLA, LC-MEC modulates the system’s point spread function (PSF) and introduces measurable intensity variations on the sensor plane under static illumination, enabling the construction of event streams from the continuously captured intensity sequence. HDR images are reconstructed by fusing the captured intensity frames and the constructed event streams using the temporal event fusion network (TEF-Net) architecture. Experiments demonstrate that LC-MEC produces structured and informative event streams in static scenes and achieves significant improvements in HDR reconstruction quality.

Directional long-borehole hydraulic fracturing is an important technique for controlling rockbursts induced by hard roofs. Its effectiveness depends primarily on whether fracturing-induced damage can modify the roof-bearing structure and thereby regulate stress concentration and elastic strain energy accumulation in the coal-rock mass ahead of the working face. However, existing numerical simulations commonly rely on predefined weakened zones or empirical parameter reduction, which makes it difficult to represent the spatial heterogeneity and mechanical evolution of rock damage during field hydraulic fracturing. Taking the 2803 goaf-side working face in Hetaoyu Coal Mine as the engineering background, this study proposes a microseismic-data-driven method for characterizing hydraulic fracturing-induced damage and incorporates it into a FLAC3D finite-difference model. The stress field, elastic strain energy field, and damage distribution ahead of the working face are compared under non-fractured and hydraulically fractured conditions. In the proposed method, the energy of fracturing-induced microseismic events is converted into the Benioff strain of numerical zones according to the attenuation law of microseismic wave propagation, and the corresponding rock damage variable is then calculated using a Weibull damage model. The fracturing-damaged rock mass is further represented by weakening the elastic modulus, cohesion, and friction angle, together with the stochastic generation of strongly damaged zones. The results show that, without hydraulic fracturing, the hard roof maintains a strong, continuous bearing capacity, resulting in a continuous lateral abutment stress concentration zone and a high elastic strain energy accumulation zone ahead of the working face and near the goaf-side boundary. After hydraulic fracturing, a patchy and locally connected high-damage weakening zone forms in the target roof strata. This damaged zone cuts the original continuous load-transfer structure through which the hard roof concentrates load toward the goaf side, reduces the extent of high-stress and high-energy zones in the coal seam, and induces an asymmetric adjustment of the dominant mining-induced energy release zone from the goaf side toward the solid-coal side. These simulation results agree well with the field observation that microseismic activity is mainly concentrated near the roadway on the solid-coal side. The study indicates that the rockburst-control mechanism of directional long-borehole hydraulic fracturing is not limited to simple overall stress dissipation. A key finding is that the fracturing-induced heterogeneous damage zone effectively interrupts the continuous load-transfer and energy-storage paths on the goaf side. This induces an asymmetric spatial redistribution of the mining-induced energy field from the goaf side toward the solid-coal side, thereby mitigating the high static-load and high-energy-storage state ahead of the working face.

Multifunctional materials including magnetism and nonlinear optical activity are of great interest due to their wide applications. Herein, we report a novel polar mixed-anion oxalate fluoride KMn(C2O4)F 1 with one-dimensional (1D) Mn2+-F-Mn2+ spin chains arranged in triangular geometry, exhibiting both 1D magnetism and nonlinear optical activity. Compound 1 crystallizes in the polar Cmc21 space group and is a nonlinear optical material with the efficiency of second harmonic generation (SHG) ∼0.2 times that of KH2PO4 (KDP). Magnetic measurement results indicate typical low-dimensional magnetism, with a broad maximum around 25 K, which suggests strong intrachain interactions with a Weiss temperature of -60(1) K. Additionally, 1 undergoes a long-range antiferromagnetic ordering at ∼11 K, which is further confirmed by the heat capacity result. Our work demonstrates the potential of using mixed anions to design and develop multifunctional materials with both low-dimensional magnetism and nonlinear optical activity.

Frequent rockbursts in staggered roadways beneath residual coal pillars pose a critical challenge for the slice mining of ultra-thick coal seams. Taking the LW250101-2 of Huating Coal Mine as a case study, this paper systematically reveals the stress evolution laws and rockburst mechanism induced by irregular residual pillars by integrating microseismic (MS) monitoring, moment tensor inversion, and numerical simulation. First, source mechanism inversion analysis elucidated that compressive-shear failure of coal pillars was the dominant rupture mode in five of the eight recorded rockburst events. Second, numerical simulations demonstrate that the width of the left wing and the thickness of the right wing of the “L-shaped” coal pillar structure are the key geometric factors controlling rockburst risk; larger dimensions correlate with more intense stress concentration and higher-energy MS events. Moreover, the stress deflection effect of “L-shaped” coal pillars causes the haulage gateway of the LW250101-2 to remain in a state of stress accumulation, increasing its susceptibility to rockburst. Finally, a synergistic prevention system consisting of deep-hole roof blasting, large-charge coal blasting, and ultra-deep large-diameter boreholes was implemented. Field monitoring confirms that these measures dissipated high-stress concentrations, reduced rockburst frequency to zero and ensured safe mining.

Single-crystal investigations are crucial for uncovering intrinsic anisotropic magnetic responses since field-induced quantum states may remain hidden in powder-averaged measurements. Herein, we report the single crystal growth and investigation of the anisotropic magnetism of the copper selenite-sulfate Cu3O2(HSeO3)(HSO4)(H2O) 1. The crystal structure of 1 contains layers of a distorted T26-type non-Archimedean lattice, where diamond chains are interconnected through dimers. The magnetic susceptibilities along the a and b axes are larger than those along the c-axis, indicating that 1 shows magnetic easy-plane anisotropy with the ab plane as the easy plane. This is further confirmed by isothermal magnetization measurements. Interestingly, high-field magnetizations along the a, b, and c axes all show 1/3 plateaus up to 35 T with magnetization of similar to 1.1 mu(B), at 8 T for H parallel to a and H parallel to b, but 10 T for H parallel to c. Long-range antiferromagnetic order is observed only at similar to 2.6 K in the chi(a)(T) data, despite the large Weiss temperature of |theta| > 300 K, indicating strong quantum spin fluctuations. This work demonstrates the occurrence of an anisotropic magnetism in non-Archimedean lattices with a robust 1/3-magnetization plateau.

Upper-slice residual coal pillars in extra-thick coal seams can form local high-stress structures and strongly affect lower-slice staggered roadways. To clarify the rockburst mechanism beneath such a pillar, this study investigated the 250101-2 lower-slice working face by integrating field damage investigation, microseismic monitoring, coal-pillar mechanical analysis, FLAC3D simulation, and destressing verification. During retreat, 5737 microseismic events (MS) were recorded, including 56 high-energy events with energy not lower than $$\:1.0\times\:{10}^{4}$$J. These events were mainly distributed within approximately 200 m ahead of the working face and concentrated on the haulage roadway side. The 20 m residual-pillar zone accounted for 41% of the high-energy event frequency and 37% of the total high-energy release, consistent with repeated field manifestations such as rib deformation, support damage, mesh failure, coal-rock collapse, and floor heave. Mechanical analysis shows that the residual pillar may behave as a narrow pillar with peak-stress superposition or evolve into a wider composite bearing structure after goaf recompaction. Numerical simulation further indicates that, during lower-slice mining, the original 20 m pillar and adjacent compacted coal-rock mass formed an effective bearing zone of approximately 54 m, enhancing stress accumulation and downward transfer. The LW250101-2 haulage roadway was located near the boundary between the high-stress zone and the stress-relieved zone, where a sharp stress gradient promoted asymmetric deformation and dynamic instability. Coordinated destressing by roof deep-hole blasting, rib deep-hole blasting, and ultra-deep large-diameter boreholes transformed microseismic activity from concentrated high-energy release to more dispersed low-energy release, reducing rockburst risk in the residual-pillar-affected roadway.

We present a portable liquid crystal (LC)-based optoelectronic hybrid neural network system for high-precision formaldehyde sensing. Central to the platform is an electrically tunable LC-on-Chip module, optimized via a progressive inverse design strategy that co-optimizes optical and neural network parameters. We introduce LC-chromatic aberration coding, a novel optical computing mechanism that efficiently captures rich spatial-spectral features, which are subsequently decoded by the integrated neural network to quantify formaldehyde with high selectivity. The compact device achieves approximately triple that of commercial kits and matches laboratory-grade spectrophotometers, despite occupying less than 1% of their volume. It further exhibits robust interference rejection against acetaldehyde and other VOCs in complex mixtures. By synergizing optical coding with co-optimized hardware and algorithm, this work bridges the gap between portability and lab-scale performance, enabling scalable, intelligent indoor air quality monitoring.

The field of see-through mixed-reality faces the challenge of accurately displaying the occlusion relationship between virtual objects and the real world. To address this issue, this study proposes an end-to-end closed-loop optimization approach using polymer-dispersed liquid crystal (PDLC) to achieve see-through, mixed-reality displays with accurate virtual and real occlusion. This study begins by constructing a PDLC light field acquisition system and deriving its optical model to obtain the system’s light intensity distribution of mixed-reality scenes. Based on the characteristics of the proposed system, an occlusion compensation network is introduced to optimize the light intensity distribution information, resulting in optimized component and occlusion network parameters. Experimental results demonstrate that the proposed method achieves accurate virtual and real occlusion display effects, particularly in low-texture environments. This study shows significant occlusion display improvements while enhancing image quality compared to traditional methods.

This study presents a liquid crystal on chip (LCOC) system featuring a tunable liquid crystal microlens array (LC-MLA) as its core optical engine, an innovative approach that tightly integrates physical optics with deep learning to resolve virtual-real occlusion inconsistencies in see-through mixed reality. By directly encoding the voltage-dependent physical point spread functions (PSFs) of the LC-MLA into convolutional kernels, the system establishes an end-to-end closed-loop optoelectronic co-processing framework. Through joint optimization of optical parameters and neural network weights via backpropagation, it accurately reconstructs depth-aware occlusion relationships without requiring auxiliary hardware such as spatial light modulators. This lightweight yet high-fidelity solution pioneers intelligent light-field manipulation for realistic augmented reality displays.

Cable routing is a common manipulation task in assembly and manufacturing, yet it remains challenging due to the deformable nature of cables and the constraints of cluttered routing environments. In this paper, we present CRAFT: Cable Routing Around Fixtures using Two grippers, a novel hardware plus software architecture that integrates unimanual and bimanual operations for long-horizon cable routing. To address jamming due to friction, we present a novel caging gripper with roller mechanism. Physical experiments consisting of 160 trials on a modified NIST board with five types of fixtures and turning angle up to 930 degrees, yield an average completion ratio of 84.5% across four routing difficulty tiers, representing a 54.2% improvement over an earlier baseline. The cable routing materials and benchmarks are available at https://manipulation-net.org/tasks/cable_routing.html.

Accurate source locating and a complete data catalogue of the seismic network are the prerequisites for seismic analysis methods to identify coal burst risks. Comprehensively understanding the spatial characteristics of source location errors and seismic data integrity is a key insight for optimising seismic networks and enhancing monitoring performance. Based on the monitored seismic data in a burst-prone longwall, this study develops two novel methodologies, Emulation-Testing-based Source Locating Accuracy Analysis (ETSLA) and Probability-based Magnitude of Completeness (PMC) method, to evaluate seismic monitoring performance in underground coal mines. The results indicate that ETSLA effectively quantifies vector characteristics of source location errors, revealing anisotropic error distributions in the studied longwall. The PMC method presents significant differences among geophones regarding their wave detection capacities. The detection probability of the seismic network for the events demonstrates progressive enhancement with increasing energy magnitude. In field practice, ETSLA can correct misclassified burst types by accounting for location errors. The Seismic data inferred using the PMC method can retrace missing seismic activity, and the inferred high-energy zones accurately correlate with actual burst damage locations. The study can serve as a reference to enhance the quality of seismic monitoring for precise early warning of coal burst risks.

Generalizable long-horizon robotic assembly requires reasoning at multiple levels of abstraction. While end-to-end imitation learning (IL) is a promising approach, it typically requires large amounts of expert demonstration data and often struggles to achieve the high precision demanded by assembly tasks. Reinforcement learning (RL) approaches, on the other hand, have shown some success in high-precision assembly, but suffer from sample inefficiency, which limits their effectiveness in long-horizon tasks. To address these challenges, we propose a hierarchical modular approach, named Adaptive Robotic Compositional Hierarchy (ARCH), which enables long-horizon, high-precision robotic assembly in contact-rich settings. ARCH employs a hierarchical planning framework, including a low-level primitive library of parameterized skills and a high-level policy. The low-level primitive library includes essential skills for assembly tasks, such as grasping and inserting. These primitives consist of both RL and model-based policies. The high-level policy, learned via IL from a handful of demonstrations, without the need for teleoperation, selects the appropriate primitive skills. We extensively evaluate our approach in simulation and on a real robotic manipulation platform. We show that ARCH generalizes well to unseen objects and outperforms baseline methods in terms of success rate and data efficiency. More details are available at: https://long-horizon-assembly.github.io.

Rockburst is a common mining hazard causing dynamic damage to coal and rock masses, posing significant threats to personnel and equipment safety. Various analytical methods exist to assess impact risks, with microseismic monitoring systems playing a pivotal role due to their stability, dynamism, and continuity. This approach utilizes a dual residual connection and a deeply connected stack architecture to facilitate seasonal-trend predictions and enhance their interpretability in time series prediction tasks using a purely deep learning model. The time-frequency and total energy of microseismic events are predicted using the proposed approach, and a comparative experimental study is conducted on the time window lengths of M = 7 days and M = 4 days. The results indicate that the proposed approach effectively predicts the evolution trend of microseismic event frequency, with minor discrepancies between the predicted results and the actual monitoring values, showing its excellent prediction performance and generalization capability.