Wireless technology is becoming increasingly prominent in industrial environments today, particularly for controlling automated guided vehicles such as mobile robots. Beyond conventional communication requirements, wireless solutions must also comply with functional safety standards such as IEC 61508 and IEC 62745. However, the susceptibility of wireless systems to RF interference and cyberattacks raises questions about their suitability for safety-critical environments. In this paper, we investigate the integration of one of the latest features of Bluetooth Low Energy, namely Broadcast Isochronous Streams (BIS), which enable efficient one-to-many broadcast communication, into industrial systems. Furthermore, we present an applicationlayer framework built on top of BIS that provides essential security and safety features such as encryption, decryption, lightweight key exchange, transmission of periodic keepalive signals, and emergency stop functionality. Experimental results demonstrate that the proposed framework adds only 33% latency overhead, while ensuring compliance with key safety and security requirements. The framework also supports network scaling to hundreds of devices and enables timely execution of safety procedures even in harsh RF conditions, demonstrating that BIS can be securely and reliably employed in industrial environments.

Sustainable transport systems are increasingly operating in contexts where uncertainty is no longer the exception but a structural condition [...]

Quantization has become a standard tool for efficient LLM deployment, especially for local inference, where models are now routinely served at 2-3 bits per parameter. The state of the art is currently split into simple scalar quantization techniques, such as GPTQ or AWQ, which are widely deployed but plateau in accuracy at 3-4 bits per parameter (bpp), and"second-generation"vector- or trellis-quantized methods, such as QTIP, GPTVQ and AQLM, which push the accuracy frontier but are notoriously hard to implement and to scale. In this paper, we ask whether this gap is fundamental, or whether a carefully optimized $\textit{scalar}$ quantizer can recover most of it. We answer in the affirmative, by introducing GSQ (Gumbel-Softmax Quantization), a post-training scalar quantization method which jointly learns the per-coordinate grid assignments and the per-group scales using a Gumbel-Softmax relaxation of the discrete grid. GSQ matches the cardinality of the relaxation to the small number of levels available in the target bit-width regime (e.g., 3-8 levels for ternary and 3 bpp, respectively), making optimization tractable. Practically, on the standard Llama-3.1-8B/70B-Instruct models, GSQ closes most of the gap between scalar quantization and the QTIP frontier at 2 and 3 bits, while using a symmetric scalar grid with group-wise quantization, and thus remains compatible with existing scalar inference kernels. We further show that the same discrete-assignment optimization can be applied to practical GGUF K-Quant checkpoints: starting from publicly released GGUF models, GSQ improves accuracy while projecting the result back into the same deployment format. Finally, GSQ scales to trillion-scale Mixture-of-Experts models such as Kimi-K2.5, where vector-quantized methods are difficult to apply. The source code is publicly available at https://github.com/IST-DASLab/GSQ.

Abusive head trauma (AHT), is considered a leading cause of fatalities resulting from physical abuse in infants under 2 years of age, with a peak incidence between 1 and 2 months after birth. The incidence of AHT ranges from 14 to approximately 40 cases per 100,000 children in industrialized countries with a mortality rate ranging from 10 to 20%. The absence of internationally recognized best practices or guidelines especially in the field of forensic medicine has resulted in methodological variability in the management of these cases across different settings. In response to this gap, a comparative working group involving experts from Italy and the Balkan countries was established, leading to the creation of a shared discussion platform. The aim of this collaborative effort was to identify strengths and critical issues in the forensic handling of abusive head trauma, ultimately with the goal of developing a shared workflow chart for the management of these complex cases within the network.

It is believed that teachers in secondary education prioritise correctness in grammar instruction and, in doing so, often overlook the functional potential of language, which is used to negotiate social meaning. The paper aims to investigate the language instruction gap in high school general English language classes by examining the extent to which lesson plans incorporate Systemic Functional Grammar (SFG) as a resource for meaning-making (Schleppegrell, 2017). The analysis in this paper applies Halliday’s (1994) metafunctional framework and Nunan’s (1995) analytical outline in order to investigate a corpus of ten lesson plans and their corresponding materials through a qualitative coding scheme. The findings reveal that the main instructions remain dominated by prescriptive rules that treat language as a static object. The analysis indicates that, despite learner-centred classroom management, the interpersonal resources of Mood and Modality remain teacher-controlled, limiting learners’ ability to actively participate in the meaning-making process. The paper concludes that SFG can contribute to bridging the instruction gap by transforming grammar from a set of restrictive rules into a dynamic, functional resource for successful communication.

The shift in agricultural production enabled by modern technology has led to the adoption of intelligent and autonomous systems, marking the Agriculture 5.0 era. Drones play a crucial role in this system, and as their use continues to expand, selecting the appropriate drones presents a complex multi-criteria decision-making problem that requires careful consideration of various factors. In this study, drones were evaluated and selected based on technical, economic, and environmental criteria. The decision-making process used interval type-2 fuzzy sets (IT2F) to incorporate uncertainty, and the model focused on ten criteria and five available drones in the Serbian market. It integrated the SiWeC (Simple Weight Calculation) method to determine criteria weights and the CRADIS (Compromise Ranking of Alternatives from Distance to Ideal Solution) method to rank the drones, forming a hybrid approach. The results indicated that Ease of Use (C6), Price (C8), and Maintenance Costs (C7) had the greatest influence on the final decision. Drone 4 achieved the highest score of 0.969 based on expert assessment, outperforming the second-ranked Drone 1 by 12%. Sensitivity analysis across 30 scenarios confirmed the stability of the ranking and showed that individual criteria did not significantly impact the results. The proposed IT2F SiWeC-CRADIS model demonstrated strong robustness and reliability and could be applied to select other smart technologies within the framework of Agriculture 5.0.

Traditional Task and Motion Planning (TAMP) systems depend on physics models for motion planning and discrete symbolic models for task planning. Although physics models are often available, symbolic models (consisting of symbolic state interpretation and action models) must be meticulously handcrafted or learned from labeled data. This process is both resource-intensive and constrains the solution to the specific domain, limiting scalability and adaptability. On the other hand, Vision Language Models (VLMs) show desirable zero-shot visual understanding (due to their extensive training on heterogeneous data), but still achieve limited planning capabilities. Therefore, integrating VLMs with classical planning for long-horizon reasoning in TAMP problems offers high potential. Recent works in this direction still lack generality and depend on hand-crafted, task-specific solutions, e.g. describing all possible objects in advance, or using symbolic action models. We propose a framework that generalizes well to unseen problem instances. The method requires only lifted predicates describing relations among objects and uses VLMs to ground them from images to obtain the symbolic state. Planning is performed with domain-independent heuristic search using goal-count and width-based heuristics, without need for action models. Symbolic search over VLM-grounded state-space outperforms direct VLM-based planning and performs on par with approaches that use a VLM-derived heuristic. This shows that domain-independent search can effectively solve problems across domains with large combinatorial state spaces. We extensively evaluate our method and achieve state-of-the-art results on the ProDG and ViPlan benchmarks.

Abstract Objectives To identify predictors of clinically inactive disease (CID) and clinical remission (CR) in patients with juvenile idiopathic arthritis receiving etanercept during the 2-year, phase 3b, open-label CLIPPER study (NCT00962741) and the 8-year extension study, CLIPPER2 (NCT01421069). Methods Patients with extended oligoarthritis (2–17 years), enthesitis-related arthritis or psoriatic arthritis (each 12–17 years) were enrolled in CLIPPER/CLIPPER2. Predictors of CID (according to Juvenile Arthritis Disease Activity Score [JADAS] and JIA-ACR response criteria) and CR (≥6 months of CID) were identified using a multivariate stepwise logistic regression model. Results Two-thirds of patients met the criteria for CID at any point and 34–43% achieved CR. Height Z-score ≥0.74, age at onset ≤12 years, normal CRP levels, HLA-B27+ status, JADAS low disease activity (LDA) at 3 months and ≤4 swollen joints were predictive of JADAS CID. BMI Z-score >0.80, age at onset ≤12 years, normal CRP levels and JADAS LDA at 3 months were predictors of JIA-ACR CID. JADAS LDA at 3 months was a predictor of JADAS CR, and height Z-score >1.23, JADAS LDA at 3 months and >12 swollen joints were identified as predictors of JIA-ACR CR. Conclusion In patients with JIA treated with etanercept, early responses to treatment in line with treat-to-target recommendations, younger age, HLA-B27+ status and lower disease activity at baseline were associated with clinically inactive disease and clinical remission. Trial registration ClinicalTrials.gov IDs: CLIPPER (NCT00962741); CLIPPER2 (NCT01421069)

Understanding behavioral differences between experimental groups and quantifying action structure in behavioral experiments remains challenging. Currently, most approaches rely on pose estimation followed by downstream classification, resulting in assay specific pipelines with substantial annotation requirements. Here we present SingleBehavior Lab (SBL), a framework for modeling behavior across experimental contexts using a standardized graphical interface. SBL leverages spatiotemporal embeddings from large video foundation models and combines them with lightweight contrastive adapters, a multi-head attention pooling (MAP) module and a temporal decoder to enable behavior sequencing and task-specific refinement. The framework supports few-shot learning, allowing small models trained on pretrained embeddings to improve action segmentation and classification with limited labeled data, without fine-tuning the underlying video model. In parallel, a large segmentation model with motion-aware memory is used to extract object-centered representations that, together with shared spatiotemporal embeddings, enable unsupervised clustering of behavioral states and analysis of their structure, including cluster prioritization, transition dynamics and attention-based interpretability. Across multiple assays and species, SBL supports identification of group-level differences and rare behaviors, and provides a basis for integrating behavioral representations across experimental contexts.

Introduction: Intestinal intussusception is the most common form of acquired intestinal obstruction in childhood. Diagnostic errors reach 50%, and complications occur in up to 53.7% of cases. Pneumoirrigoscopy under fluoroscopic control (PIS) has been the standard conservative technique, but its radiation burden and single-attempt limitation prompted the development of ultrasound-guided hydroechocolonographic disinvagination (HEC). This study aimed to compare the clinical outcomes of HEC and PIS in paediatric intestinal intussusception. Methods: A retrospective comparative study of 132 children aged 2 months to 10 years treated at the Specialized Paediatric Surgical Clinic of Samarkand State Medical University between 2000 and 2023. The control group (CG, n=59; January 2000–December 2013) received conventional PIS; the study group (SG, n=73; January 2014–December 2023) underwent ultrasound-guided HEC, a radiation-free technique developed at our institution. Primary outcomes were the rate of successful conservative reduction (assessed per patient), length of hospital stay, and mortality. Chi-squared and Student’s t-tests were used; p<0.05 was considered significant. Results: The predominant age group was 6 months to 1 year (51.5%); males predominated (70.5%, p<0.001). The ileocaecal variant was found in 90.1% of patients. Successful conservative reduction was achieved in 53.4% (39/73) of the study group versus 39.0% (23/59) in the control group. Mean hospital stay was significantly shorter in the study group (2.5±0.66 vs 4.6±0.51 days, p<0.05). Group-specific mortality was 2.7% (2/73) in the SG versus 8.5% (5/59) in the CG. Conclusion: Ultrasound-guided HEC is a safe, effective, and radiation-free alternative to PIS, achieving a higher rate of conservative reduction, a significantly shorter hospital stay, and enabling multiple reduction attempts. Disease duration alone should not determine treatment modality; clinical condition and the absence of peritoneal signs must be considered jointly.

The relativizer kamā, a compound consisting of Arabic preposition ka and the nominal relative pronoun mā, is a polyfunctional expression used very frequently in Modern Standard Arabic. Its description in the literature, however, remains incomplete, unsystematic and mostly morphologically motivated. The syntactic functions of the relativizer kamā are therein mentioned only rarely and sporadically, lacking a systematic analysis. Therefore, the main goal of this paper is the analysis of various syntactic functions that the relativizer kamā and clauses it introduces can take both within the structure of the main clause and outside its structure, at the text level. The analysis is based on the analyticaldescriptive method and the typological-functional approach. Its results show that clauses introduced by the relativizer kamā in Modern Standard Arabic cover a wide spectrum of syntactic functions, ranging from comparative clauses, both factual and hypothetical, via attributive, predicate and verb complement clauses, to the functions of sentence modifying adverbial and connector at the text level. In addition to providing a systematic description of the various functions of the relativizer kamā and constructions introduced by it, the analysis presented in the paper also draws attention to stylistic efficiency of such constructions and the important role they play in enriching the inventory of means of stylistic choice in Arabic.

Outcome improvement alone does not reveal whether users actually inspected disclosed evidence or simply followed a highlighted recommendation. This companion human-study paper analyzes model-selection deliberation under a staged multi-criteria disclosure interface for educational quality assurance (QA). In the final filtered analytic sample of 38 participants and 228 completed scenarios, we examine interface telemetry, participant-level self-report, acceptance, task-level heterogeneity, and a heuristic low-engagement robustness check. Nonparametric comparisons and a clustering-adjusted GEE model were used. Disclosure uptake was selective: ranking was used in 60.5% of scenarios, weights in 47.4%, heatmap in 46.9%, and textual interpretations in 41.7%. Self-reports aligned with telemetry for four of the five major components. Improved scenarios showed longer Step 2 deliberation and greater engagement with multiple evidence surfaces, while the GEE model indicated that ranking use was significantly associated with improvement. Acceptance was favorable (34/38 preferred the agent-assisted workflow), whereas a low-engagement subgroup showed sharply reduced benefit. The observed pattern is more consistent with structured multi-surface deliberation than with shallow recommendation following.

The influence of infill density, number of layers, and fibre type on the tensile mechanical properties of composite parts produced by FDM 3D printing with continuous fibre reinforcement (CFR) was investigated. Specimens made of ONYX composite material reinforced with carbon, glass, and aramid fibres were tested using a static tensile test according to ISO 527. A linear regression model was developed to correlate mechanical properties with 3D printing parameters. However, the influence of infill density could not be reliably determined due to the automatic generation of solid infill around fibre layers and at boundary layers in the utilized 3D printing software. These reinforcements provided varying degrees of enhancement, with carbon and glass fibres showing the highest increase in strength, glass fibres offering the best enhancement in fracture strain, and carbon fibres in stiffness. The obtained models for tensile strength, strain at maximum stress, and modulus of elasticity can be useful in the design of 3D printed parts, offering a simple solution for the prediction of mechanical properties.