IntroductionDisaster-response robots and other embodied agents can use social-media imagery as contextual evidence, but existing approaches typically analyze affect at the image level or model diffusion without isolating affect-specific information.MethodsWe encoded disaster images into 128-dimensional continuous visual-affect embeddings representing visual intensity, fear, sadness, and negative valence. These embeddings initialized node states in a time-aware graph neural network whose edges were constrained by observed interaction links, temporal precedence, and spatial reachability. Multi-task heads predicted propagation scale, propagation speed, and spatial-ranking consistency, and a parameter-decomposition module separated affect-associated contributions from structure-associated residuals. Experiments used 4,450 reconstructed image-centered CrisisMMD samples, a common 70/15/15 split, five random seeds, matched baselines, and ablation tests.ResultsThe full model achieved an AUC of 0.842 ± 0.011, an RMSE of 0.091 ± 0.007 under the current task definition, and a Kendall tau of 0.730 ± 0.018. Relative to matched ablations, visual affect increased AUC by 0.041, time-aware recurrence reduced RMSE by 33.6%, and spatial constraints increased Kendall tau by 0.089. The RMSE unit must be standardized after the propagation-speed versus propagation-time definition is resolved, as marked in the proof.DiscussionVisual affect provides an interpretable complementary signal for modeling disaster-image diffusion. The findings do not establish causal effects, cross-platform generalizability, real-time device performance, or closed-loop robot benefit; the framework should therefore be regarded as an upstream social-sensing prior for embodied decision support.
BackgroundWearable lower-limb exoskeletons have the potential to support intention-driven control of lower-limb exoskeletons, but existing control strategies often rely on mechanical or manual triggers that fail to capture user intent.MethodSubjects were recruited from 23/9/2024 to 10/2/2025. A BiLSTM detector was pretrained on a dataset collected from 50 healthy volunteers (45 for training, 5 for independent testing) using bilateral surface EMG recordings from six lower-limb muscles (12 EMG channels), 16-channel EEG, and hip–knee kinematics. Seven naive participants then completed ten 20-m outward-and-return walking trials (10 m outward and 10 m return) under each of three control modes (EMG, EEG, hybrid). Primary outcomes were triggering latency and classification accuracy. Triggering latency was defined as the time interval between the onset of the gait-transition event and the activation of the exoskeleton assistance command. This latency included the observation delay introduced by the sliding window, feature extraction time, BiLSTM inference time, and communication delay between the decoder and the exoskeleton controller. Usability was assessed with donning/doffing times and QUEST 2.0.ResultsThe hybrid BiLSTM detector achieved higher classification accuracy (left: 91.3%; right: 86.6%) and shorter mean per-step triggering latency (left: 0.29 s; right: 0.28 s) than either EMG-only (mean 0.33 s) or EEG-only approaches (mean 0.30 s). Hybrid EEG–EMG fusion therefore improved decoding performance while reducing triggering latency compared with unimodal decoding strategies. The latency reduction relative to EMG-only control corresponded to a large effect size (Cohen’s d ≈ 0.88). Hybrid sessions also yielded shorter total session time (mean 32.1 min). Usability metrics demonstrated acceptable donning/doffing times and favorable QUEST 2.0 scores (mean 32.0/40).ConclusionThese proof-of-concept results demonstrate that BiLSTM-based fusion of EEG and EMG improves responsiveness and classification reliability for exoskeleton assistance. These findings underscore the contribution of EEG-derived sensorimotor features and EMG information for intention-related gait transition detection within a multimodal real-time exoskeleton control framework. We discuss limitations related to sample size, artifact validation, and generalizability and identify next steps for patient studies and ergonomic optimization.
IntroductionLearning robust visuomotor policies for bimanual manipulation remains challenging due to the stringent requirements for precise coordination between arms and the ability to generalize across diverse environmental conditions. Existing diffusion-based policies often suffer from temporally inconsistent action generation, while their reliance on sparse point cloud representations limits structural completeness and fails to capture future scene dynamics, hindering performance in long-horizon tasks.MethodsTo address these limitations, we introduce GSP-3D, a unified framework that integrates generalizable 3D Gaussian Splatting (3DGS) with diffusion-based policy learning. GSP-3D comprises three key components: (1) a Generalizable Gaussian Regressor (GGR) that predicts 3D Gaussian primitives from a single RGB-D frame in real time; (2) a 3DGS-conditioned diffusion policy that aggregates Gaussians into compact latent representations, replacing point clouds with explicit geometric primitives; and (3) a transformer-based world model that forecasts future Gaussian sets and leverages prediction error as an auxiliary loss to enforce temporal consistency across actions.ResultsWe evaluate GSP-3D on the RoboTwin 2.0 benchmark across a range of bimanual manipulation tasks under both clean and domain-randomized conditions, including variations in lighting, backgrounds, and tabletop distractors. Experimental results show that GSP-3D consistently outperforms existing baseline methods, achieving higher success rates while maintaining minimal computational overhead.DiscussionThese findings demonstrate that integrating explicit 3D Gaussian representations with diffusion policies offers an efficient and robust solution for long-horizon, temporally coherent bimanual manipulation, effectively addressing the generalization and consistency challenges that limit current approaches.
Brain–computer interfaces (BCIs) are among the most prominent communication technologies that establish a direct channel for information exchange between the brain and external devices. They have been extensively applied in the field of aerospace. However, traditional BCI technology faces challenges, including a limited number of brain control commands and insufficient recognition accuracy in electroencephalography (EEG) decoding. These limitations make it difficult for traditional BCIs to perform complex tasks with high accuracy. Therefore, this study proposed a novel group BCI (G-BCI) system and further constructed a brain–machine shared control method for unmanned aerial vehicle (UAV) swarm control. First, a novel G-BCI paradigm combining precise hand movements and visual evoked potentials was designed. Moreover, an improved multi-domain feature fusion convolutional neural network (MDFF-CNN) was employed to decode EEG and electromyography (EMG) signals from precise hand movements, while a Filter Bank Common Spatial Patterns with Canonical Correlation Analysis (FBCCA) method was used for Steady-State Visual Evoked Potentials (SSVEP) decoding. Furthermore, a task-driven shared control model mapping the G-BCI system and the leader–follower UAV swarm control strategy was proposed. To verify the effectiveness of the proposed method, eight participants were recruited to conduct both offline and online experiments. The proposed G-BCI system achieved an offline accuracy of 88.91 ± 5.06% and an online accuracy of 88.89 ± 1.96%. All the experimental results demonstrate the feasibility of the proposed method.
Photovoltaic power generation systems exhibit multi-peak power–voltage characteristics under partial shading conditions, severely limiting the effectiveness of conventional maximum power point tracking methods. This paper proposes a two-layer hierarchical MPPT architecture based on an adaptive fuzzy-weighted eagle perching optimization algorithm, designated AF-EPO. Unlike existing fuzzy-metaheuristic MPPT hybrids that apply fuzzy logic as an exogenous regulator to secondary control variables while retaining fixed-parameter core dynamics, AF-EPO embeds the fuzzy inference system directly into the endogenous EPO scaling factor, jointly driven by iteration progress and population diversity, thereby addressing three persistent limitations of metaheuristic MPPT through structural changes absent from existing fuzzy-hybrid formulations: the fixed-parameter exploration–exploitation dilemma, the computational overhead of unconditional global search, and the residual steady-state power oscillation that persists in all population-based methods. A slope sign-reversal detection module first identifies whether the power–voltage curve is unimodal or multi-modal, activating the global search layer only when partial shading is confirmed. In the global layer, a 25-rule Mamdani fuzzy inference system dynamically adjusts the EPO scaling factor according to iteration progress and population diversity, balancing exploration and exploitation throughout the search. A variable-step incremental conductance controller then refines the operating point to suppress steady-state oscillation. Simulations across three scenarios demonstrate that AF-EPO reduces tracking time by 47.3%, power oscillation rate by 74.6%, and energy loss by 38.2% compared with standard EPO. Hardware experiments on a TMS320F28335 DSP platform confirm tracking efficiencies exceeding 98.1%, with a maximum simulation-to-experiment deviation of 1.1%, validating the practical effectiveness of the proposed method.
Introduction:Tripping is a leading cause of falls, and Minimum Foot Clearance (MFC)-the lowest vertical swing-foot displacement during mid-swing-is a key gait event associated with tripping risk. Preventing tripping requires not only sufficiently high MFC but also consistent control across steps, which depends on neuromotor rather than strength-based mechanisms. We hypothesized that aligning movement intention with repeated, optimal ankle motion using a surface electromyography (sEMG)-driven exoskeleton would improve neuromotor consistency and MFC characteristics. Methods:A total of 12 healthy adults completed 5 min of treadmill walking at 4 km/h before and after a 5-min ankle exoskeleton training sequence ('feet-flat, toes-up, feet-flat, heels-up', 60 bpm). The Vicon motion capture system tracked the reflective markers placed on the heel and toe to collect MFC dataset. The exoskeleton (i.e., Hybrid Assistive Limb) detects voluntary neural drive through sEMG and converts it into assisted ankle motion, supporting intention-based neuromotor training. Results:The obtained MFC dataset showed that training increased the lowest percentile of the MFC distribution (i.e., the risky tail) by 0.39 cm, reduced step-to-step variability, and increased dataset skewness, suggesting reduction of both low and high extremes in swing-foot clearance. Discussion:These preliminary findings suggest that brief intention-based ankle exoskeleton training may acutely improve safety-relevant characteristics of MFC control. Further studies in older adults and neurological populations are required to determine clinical relevance for tripping-risk reduction.
Quantifying motion similarity, despite its inherently subjective nature, is a foundational problem for humanoid motion transfer and control across different embodiments. Drawing from cognitive studies revealing that human motion similarity judgments are strongly influenced by spatial relationships among body parts and proximal contacts, we develop a graph-based representation that effectively encapsulates both features. This representation enables the definition of a robust similarity metric through graph distance computations. The proposed metric emphasizes spatial, especially proximal, relationships between body parts, facilitating motion retargeting that preserves these perceptually motivated relational cues across humanoid embodiments with shared semantic body parts and varying Degree of Freedom (DoF) configurations and body proportions. For quantitative retargeting evaluation, we introduce an order-preserving spatial similarity metric that measures how consistently inter-joint distance rankings are preserved between source and target motions.
Quadruped robots have attracted increasing attention because they can traverse uneven terrain, support field deployment, and perform tasks that are difficult for wheeled or tracked platforms. Recent advances in artificial intelligence (AI) have further expanded their capabilities from manually designed gait control toward learning-based locomotion, perception-aware adaptation, dynamic motion skills, autonomous recovery, manipulation, energy-aware operation, fault diagnosis, and human–robot interaction. However, the literature on AI-driven quadruped robotics is distributed across diverse technical topics, robot platforms, validation settings, and performance metrics, making it difficult to assess the maturity and practical value of different approaches. To address this need, this review provides an AI-centered and deployment-oriented overview of quadruped robotics. A systematic literature search was conducted using Web of Science, IEEE Xplore, ACM Digital Library, ScienceDirect, and SpringerLink, covering studies published approximately from 2000 to 2025. After screening and eligibility assessment, 287 studies were included for detailed review. The review first examines AI-driven locomotion, including reinforcement learning, non-RL machine-learning methods, model-based approaches, and hybrid strategies, with attention to robustness, sim-to-real transfer, sensor use, computational requirements, and hardware validation. It then summarizes AI-supported advanced behaviors, including jumping, fall prevention and recovery, and object manipulation, focusing on reported quantitative performance, impact management, and reliability. Finally, it discusses system-level topics that affect real-world deployment, including fault diagnosis, energy-efficient control, shared autonomy, trust-aware and explainable interaction, and safety-aware human–robot collaboration. By organizing the literature according to robot capabilities, validation maturity, and deployment challenges, this review helps clarify the current progress, limitations, and future directions of AI-driven quadruped robots.
IntroductionAccurate gait recognition using wearable sensors is of significant clinical value for adaptive prosthetic control, lower-limb exoskeleton assistance, and objective rehabilitation assessment. However, subject-independent recognition remains a major challenge, as unseen individuals can exhibit highly variable limb kinematics and muscle activation patterns, and existing approaches often rely on a single sensor modality or naive fusion strategies that fail to leverage the complementary information between inertial and electromyographic signals.MethodsTo address these gaps, this study proposes DSAF, a dual-stream attention fusion network that separately encodes kinematic (IMU) and neuromuscular (EMG) information, and adaptively integrates them through a physiological complementarity weighting mechanism designed for window-level modality adaptation. The framework is evaluated on the public HuGaDB dataset for eight common locomotion activities (e.g., walking, running, stair negotiation, and sit-to-stand transitions), using a leave-one-subject-out protocol to rigorously assess generalization to new users.ResultsDSAF achieves 96.41% accuracy, 96.08% macro-precision, 95.62% macro-recall, and 95.81% macro-F1, consistently outperforming recent sequence-learning baselines across all 18 held-out subjects. Ablation studies further confirm that both the modality-specific encoding and the adaptive fusion mechanism contribute positively to the performance.DiscussionThese findings indicate that adaptive IMU-EMG fusion can effectively strengthen wearable gait recognition, providing a promising solution for real-world rehabilitation monitoring and assistive human-machine interfaces.
Background:Spinal cord injury (SCI) leads to significant motor and sensory impairments, reducing independence and quality of life. Conventional rehabilitation often lacks the intensity and task specificity required for optimal gait recovery. Powered lower-limb exoskeletons have emerged as a promising adjunct to enhance functional outcomes through repetitive, task-oriented training. Objective:This meta-analysis aimed to evaluate the effectiveness and safety of exoskeleton-assisted rehabilitation compared with conventional therapy in improving walking ability, motor function, and activities of daily living in individuals with SCI. Methods:A systematic search of PubMed, Embase, Cochrane Library, Scopus, Web of Science, CNKI, Wanfang, and VIP databases was conducted for studies published between January 2016 and December 2024. Ten controlled trials involving 412 participants were included. Primary outcomes were Berg Balance Scale (BBS), 6-Minute Walk Distance (6MWD), Lower Extremity Motor Score (LEMS), Walking Index for Spinal Cord Injury II (WISCI-II) and Modified Barthel Index (MBI). Data were pooled using fixed- or random-effects models based on heterogeneity. Results:Exoskeleton-assisted rehabilitation significantly improved balance (BBS: MD 4.84, 95% CI 4.19-5.50; p < 0.001), walking endurance (6MWD: MD 31.09 m, 95% CI 27.25-34.93; p < 0.001), LEMS (MD 10.00, 95% CI 8.31-11.69; p < 0.001), gait independence (WISCI-II: MD 3.25, 95% CI 2.65-3.85; p < 0.001), and activities of daily living (MBI: MD 3.44, 95% CI 1.23-5.65; p = 0.003). Improvements were also observed in the ASIA lower-limb motor score (MD 7.28, 95% CI 6.44-8.12; p < 0.001). Substantial heterogeneity was present in some outcomes. No serious adverse events related to exoskeleton use were reported across the included studies. Conclusion:Exoskeleton-assisted rehabilitation is associated with statistically significant improvements in balance, walking performance, motor function, and functional independence in individuals with SCI compared with conventional therapy.
Drawing inspiration from the brain's neurocognitive mechanisms of information chunking and topographic mapping, adaptive decision-making requires neural-grounded architectures that are interpretable and resilient to uncertainty. In this paper, we propose a novel boosted fuzzy manifold granule hypersurface classifier (BFMGHC). The algorithm performs classification at the “information granule" level, realizing an intelligent modeling method that is closer to human cognition, more interpretable, and more accommodating of uncertainty. The classifier mainly consists of three main parts: (1) A manifold-based measurement method for samples that preserves local topological structure is designed, echoing the topographic representations in neural dynamics. Based on this, a global optimization clustering algorithm is proposed and integrated with the Dask framework to achieve scalable hierarchical parallel granulation from raw inputs to high-level semantic granules. (2) In the fuzzy manifold granule space, a measurement method and a hypersurface classifier are constructed, utilizing a particle swarm optimization method for parameter solving. (3) To improve interpretability, weights are assigned to different granules and base classifiers, resembling bio-inspired neuromodulation to ensure stable behavior. The proposed BFMGHC was verified on three financial risk assessment datasets in the UCI Machine Learning Repository (Default of Credit Card Clients, Bank Marketing, and German Credit Data) and achieved superior performance.
Optical flow estimation is a low-level module in computer vision, widely used in tasks such as visual odometry, autonomous driving, high dynamic range (HDR) imaging, and action recognition. Existing event-based optical flow estimation approaches suffer from scarcity of dense real-world datasets, while some unsupervised frameworks have reduced reliance on large-scale datasets by forward-backward consistency loss, they primarily exploit a 1D temporal reversal, while largely ignoring the rotational and scaling motions ubiquitous in robotics and automotive scenes. This work introduces radial consistency, a self-supervised pre-training framework that maps the event stream to log-polar coordinates and tessellates the spatial domain into K radial rings and L angular sectors, a shared encoder-decoder to predict four complementary flow fields whose cyclic sum is driven to zero, yielding a closed-loop constraint that generalizes the classical forward-backward check to 360° within a sector. Our core contribution, the radial consistency loss, is completely label-free, together with auxiliary terms, enabling self-supervised pre-training on large-scale event data. We optionally apply supervised fine-tuning on small labeled sets to adapt to specific domains, achieving competitive accuracy with fully supervised methods. Validation experiments on Multi Vehicle Stereo Event Camera (MVSEC) dataset demonstrate strong performance: our method achieves 0.67 EPE averaged across all sequences, surpassing E-RAFT (0.89 EPE) and EV-FlowNet (1.10 EPE), without any additional data. On the rotation-heavy indoor_flying3 sequence specifically, we achieve 0.93 EPE (fine-tuned) and 1.49 EPE (self-supervised only) vs. E-RAFT 1.66. We also improve upon E-RAFT in computational efficiency [55 frames per second (FPS) and 26 giga floating-point operations (GFLOPs) vs. 42 FPS and 38 GFLOPs], while requiring only minimal supervised fine-tuning.
IntroductionRGB-T salient object detection remains challenging because visible and thermal features often show structural shifts, weak thermal boundaries, and background interference.MethodsThis study proposes PSRNet, a phase-guided frequency-domain structure reconstruction network. RGB and thermal features are decomposed into amplitude and latent-phase components in the Fourier domain. Reliable cross-modal structural cues are aligned through amplitude-weighted phase consistency, and boundary-related high-frequency responses are reconstructed with a bounded Gaussian high-pass gate before adaptive phase-modulated fusion and multi-scale supervision.ResultsOn VT1000, PSRNet achieved an S-measure of 0.903, an MAE of 0.028, and an F-measure of 0.794 at threshold 0.80, with a boundary IoU of 0.862.DiscussionThe results indicate improved structural preservation and boundary recovery under challenging RGB-T conditions.
Large-scale outdoor robot navigation increasingly demands SLAM systems capable of operating efficiently across diverse and challenging terrain. While single-robot approaches face inherent coverage and computational limitations, distributed multi-robot frameworks extend this capability through collaborative mapping—yet they still degrade in complex outdoor environments due to two unresolved challenges: redundant ground points in raw point clouds overload feature extraction and loop-closure matching, while fixed ground segmentation thresholds fail on sloped terrain causing misclassification and trajectory degradation. We address the first challenge by integrating ground segmentation preprocessing as a parallel stage for each robot within the distributed SLAM framework, reducing point cloud size by 50.78% and achieving a 21.4% RMSE improvement for Robot 0 (7.99 m → 6.28 m) compared to the unprocessed baseline. We address the second challenge with the proposed SAGS (Slope-Adaptive Ground Segmentation) module, which continuously monitors platform tilt via IMU orientation and dynamically interpolates ground segmentation parameters within a 5°–15° tilt range; SAGS recovers Robot 1 RMSE from 8.48 m to 6.23 m (26.5% improvement) on sloped terrain without flat-terrain penalty (GPS-validated 1.083 m RMSE on a public 612 m benchmark). Both contributions are validated through progressive three-stage ablation evaluation on a campus three-robot dataset (heterogeneous team: two wheeled ground robots and one legged quadruped, diverse terrain including sloped sections) and cross-validated on a public GPS benchmark (612 m, GPS ground truth), confirming the independent contribution of each system component.
Recently, UAV path planning in 3D complex environments has attracted increasing attention due to its significance in UAV motion control systems. However, the NP-hard nature of this problem poses significant challenges in generating a high-quality path. To address this issue, this paper proposes an improved self-adaptive particle swarm optimization (ISAPSO) algorithm by integrating the standard PSO 2011 with evolutionary game theory (EGT). Firstly, a novel self-adaptive parameter updating strategy is proposed, which combines the evolutionary stable strategy in EGT with hyperbolic tangent function to balance the exploration and exploitation capabilities of ISAPSO. Subsequently, an ISAPSO-based path planning approach is developed to generate optimal 3D path for UAV in an obstacle-rich environment. To efficiently handle constraints, a novel self-adaptive constraint handling technology is proposed in the developed path planner. Finally, the performance of the proposed ISAPSO is evaluated against six state-of-the-art evolutionary algorithms using 20 test functions. Following the benchmark study, the ISAPSO-based path planner is is validated in different scenarios against six well-known counterparts. The simulation results confirm that the proposed ISPASO outperforms its competitors in the benchmark study at a 90% confidence level. Moreover, the ISPASO-based path planning method dominates its contenders in terms of the path optimality. Therefore, the proposed method could be regarded as a vital alternative in the area of path planning.
Blue cheeses owe their distinctive texture, flavor, and aroma to Penicillium roqueforti. Understanding the technological traits and secondary metabolite production of this species is essential for cheese quality and safety. Here, 20 P. roqueforti isolates from traditional Turkish blue cheeses, including Tulum and Civil, were evaluated for growth at different temperatures, salt tolerance, proteolytic and lipolytic activities, and production of mycophenolic acid (MPA) and roquefortine C (ROQC). Marked strain-level variation was observed. Hierarchical clustering and principal component analysis grouped the isolates into three clusters. Civil cheese isolates showed improved growth under temperature and salt stress and produced lower ROQC than Tulum isolates, suggesting adaptation to distinct cheese environments. Selected isolates were tested in model Tulum cheeses, where all successfully colonized and formed blue veins. Secondary metabolite levels in cheese were low. These results highlight the diversity of Turkish P. roqueforti isolates and support the development of cheese-specific starter cultures.
This study characterized a novel extracellular acid protease from Lacticaseibacillus paracasei (APLP) and evaluated its potential as a milk-clotting enzyme for cheese making. APLP was purified by heat treatment, ammonium sulfate precipitation, and size-exclusion chromatography, yielding a 30 kDa enzyme that showed effective milk-clotting activity at pH 6 and 35 degrees C, despite optimal proteolytic activity at pH 8.5 and 50-75 degrees C. Under equal-volume conditions (1 mL enzyme extract per 10 mL milk), APLP achieved flocculation in 71 s, substantially faster than the commercial comparator Presurpara 1/5000 (484 s). Coagulation activities were 13.9 PU/mL versus 2.04 PU/mL, respectively. APLP also exhibited superior specific activity (8.74 PU/mg) and coagulation power (1/42,810) compared to Presurpara 1/5000 (1.17 PU/mg and 1/25,196, respectively). A prototype cheese (Lacticop) produced with APLP and enriched with rosemary received favorable sensory evaluation scores. These findings indicate that APLP is a promising microbially derived coagulant for large-scale cheese production.
Calcium carbonate (CaCO3) addition to milk represents a promising approach to reduce phosphorus bioavailability in cheese for consumers with impaired renal function, but no data is available on the structural, sensory characteristics of cheese. This study evaluated the effects of CaCO3 supplementation (2 g/L) on physico-chemical, structural, sensory properties of Caciotta cheese. CaCO3 significantly increased cheese porosity, likely due to CO2 release under mildly acidic conditions, while textural and rheological properties were unaffected. Low-field NMR relaxometry revealed reduced proton relaxation times in specific water populations, suggesting altered water-protein interactions for CaCO3-enriched cheese. Sensory analysis indicated slightly lower but positive consumer acceptability scores for CaCO3-enriched cheese, mainly related to appearance and flavor, whereas texture perception remained unchanged. Overall, CaCO3 supplementation enabled the production of nutritionally functional Caciotta cheese preserving satisfactory structural and sensory quality. These findings provide useful insights into the development of dairy products targeted at special nutritional needs.
The current study aims to understand the contribution of koku-active substances such as gamma-glutamyl peptides and volatile compounds to the overall perception of koku-related sensory properties using multivariate statistical analysis. Specifically, we examined blue-mould, smear-ripened and hard yellow type cheeses and cheese powders. The results highlighted blue-mould cheese and cheese powder among other cheese types, showing the highest diversity and quantity of gamma-glutamyl peptides. Among the seven different measured kokumi peptides, gamma-Glu-Thr and gamma-Glu-Glu showed the strongest correlations with koku-related descriptors such as mouthfulness, richness, and persistence of aftertaste. Volatile markers, including esters, alcohols, terpenes, and sulfur compounds, were strongly linked to sensory attributes typically associated with smear- and blue-mould-type cheeses, such as 'smear flavour', 'blue cheese culture', or 'mouldy'. These findings confirm the synergistic role of gamma-glutamyl peptides and volatiles in shaping flavour complexity and koku perception. The proposed multivariate framework offers a robust tool for guiding formulation strategies in cheese-based products.
This study investigated the surface characteristics and the rehydration behavior of sodium caseinate powder obtained from camel and bovine milk (CMSCP, BMSCP, respectively). Both powders exhibited similar total fat and ash contents, whereas CMSCP contained slightly lower protein and higher lactose quantity than BMSCP. Despite having comparable fat content, the X-ray photoelectron spectroscopy indicated that CMSCP exhibited greater surface fat (48.9 +/- 2.4%) and lower protein (50.4 +/- 2.5%) contents than BMSCP. This highlighted the hydrophobic nature of CMSCP surface which was confirmed by its higher C/O ratio (5.6 +/- 0.3). FT-IR spectroscopy confirmed the structural differences between both powders. CMSCP displayed strong hydrophobic protein-protein interaction and a higher beta-sheet structure, as well as a unique glycerol-casein band at 1037 cm-1, indicating the development of insoluble aggregates. Such protein aggregations, surface fat coverage and surface hydrophobicity deeply reduced the rehydration behavior of CMSCP which was confirmed by the turbidity measurements.
