Medrobotics (United States)

The intrinsic soft nature of compliant supernumerary limbs and exosuits makes them appealing candidates for assisting human movements, with potential applications in healthcare, human augmentation, and logistics. In the following chapter, we describe the technology used in exosuits and supernumerary limbs for assistance with activities of daily living, with emphasis on aiding grasping and flexion/extension of the elbow joint. We discuss the mechanical design principles of such devices, detail the control paradigms that can be used for intention-detection, and present the design and evaluation of cutaneous interfaces used for force feedback rendering. Tests on healthy and impaired subjects highlight that exosuits and supernumerary limbs are potential cost-effective and intrinsically safe solutions for increasing the capabilities of healthy subjects and improving the quality of life of subjects suffering from motor disorders.

Inspection robots are imperative for monitoring industrial machinery like belt conveyor systems in mines and steel mills. However, gathering sufficient training data for developing an effective inspection algorithm poses a challenge. In environments characterized by safety concerns and diverse conditions, audio or vibration signals or both often prove more reliable for fault detection than visual data. However, practical data collection remains difficult due to safety, security, and environmental constraints. This study addresses the training data scarcity by exploring publicly available datasets. The likelihood of encountering novel, untrained fault sounds in real-world scenarios was considered while introducing a novel sequence-to-sequence (Seq2Seq)-based deep neural network model capable of identifying such abnormalities, particularly in belt conveyor systems. Our model aims to discern whether unfamiliar sounds are typical or anomalous, acknowledging the inherent complexity of precisely classifying unknown sounds. The training involved public datasets comprising two main categories: normal and abnormal sounds, each encompassing nine distinct types. Subsequently, we evaluated our model using separate datasets entirely distinct from the training data. Our proposed model achieved an overall accuracy of 87.53 % with untrained datasets, including four environmental sounds (two usual, two unusual) and two belt conveyor operational noises (normal and faulty). Furthermore, we implemented our algorithm in a wire-guided moving robot, ConveyorGaze, tailored for inspecting belt conveyor systems in steel mills, demonstrating the feasibility of our approach.

The 3D weakly-supervised visual grounding task aims to localize oriented 3D boxes in point clouds based on natural language descriptions without requiring annotations to guide model learning. This setting presents two primary challenges: category-level ambiguity and instance-level complexity. Category-level ambiguity arises from representing objects of fine-grained categories in a highly sparse point cloud format, making category distinction challenging. Instance-level complexity stems from multiple instances of the same category coexisting in a scene, leading to distractions during grounding. To address these challenges, we propose a novel weaklysupervised grounding approach that explicitly differentiates between categories and instances. In the category-level branch, we utilize extensive category knowledge from a pre-trained external detector to align object proposal features with sentencelevel category features, thereby enhancing category awareness. In the instance-level branch, we utilize spatial relationship descriptions from language queries to refine object proposal features, ensuring clear differentiation among objects. These designs enable our model to accurately identify target-category objects while distinguishing instances within the same category. Compared to previous methods, our approach achieves state-of-the-art performance on three widely used benchmarks: Nr3D, Sr3D, and ScanRef.

Abstract Higher expectations for safety, well integrity and Coiled Tubing (CT) string durability have resulted in increasing demand for periodic inspections and continuous monitoring of CT strings. As such, leading companies in the Middle East are requiring more stringent inspections. In this paper, we outline the challenges presented by stricter inspection requirements amid the recent trend for heavier walled strings. We also discuss the engineering solutions we employ to address these new challenges and share concrete results that proved effective. The new inspection requirements go beyond the stringent API-5ST-SR37 specification for new CT strings. The trend towards heavier wall thickness in CT strings makes the new specifications even more challenging as this increases the separation between the external inspection sensors and any internal flaws. This separation degrades the sensitivity of the inspection tools to internal flaws. To address these challenges, we revamped our CT inspection and continuous monitoring solutions. Specifically, we have developed the following solutions: A new software version for our ECTAS III inspection system with advanced signal processing algorithms that improve sensitivity to difficult flaws, such as narrow longitudinally oriented mass losses on the outer or inner tubing surface.Significant hardware and software enhancements to our continuous, real-time CT monitoring system. These enhancements yielded significantly better results as proven by data collected upon scanning third-party certified reference standards. These standards come from CT samples supplied by various mills, are EDM machined with reference flaws and independently certified according to industry specifications that we aim to comply with. We share the performance results of the scans through plots depicting the Magnetic Flux Leakage (MFL) response of the reference flaws. The results show an increase in tool sensitivity to flaws evidenced by a minimum increase of 63% in the signal to noise ratio (SNR) across key reference flaws. This increase in the SNR depends on the wall thickness of the reference standard as well as the flaw geometry. Our results show that our probability of detection (POD) has improved dramatically. We are now for the first time able to detect the most difficult reference flaw of the new specifications: the longitudinal inner diameter (LID) flaw. The results highlight the unprecedented achievement of a minimum SNR of 3 to 1 on LID flaws in tubing 2 3/8 inch in diameter and up to 0.276 inch in wall thickness. Our technological improvements have ensured continuation in meeting industry requirements amid challenging changes in CT specifications. We continue to contribute to leading CT operators’ endeavor to ensure higher CT quality, enhance safety and drive operational excellence.

Forest structure has a strong relationship with abiotic components of the environment. For example, canopy morphology controls snow depth through interception and modifies incoming thermal radiation. In turn, snow water availability affects forest growth, carbon sequestration, and nutrient cycling. We investigated how structural diversity and topography affect snow depth patterns across scales. The study site, Grand Mesa, Colorado, is representative of many areas worldwide where declining snowpack and its consequences for forest ecosystems are increasingly an environmental concern. On the basis of a convolution neural network model (R2 of 0.64; root mean squared error of 0.13 m), we found that forest structural and topographic metrics from airborne light detection and ranging (lidar) at fine scales significantly influence snow depth during the accumulation season. Moreover, complex vertically arranged foliage intercepts more snow and results in shallower snow depths below the canopy. Assessing forest structural controls on snow distribution and depth will aid efforts to improve understanding of the ecological and hydrological impacts of changing snow patterns.

Introduction: Glucose Hydrogen Breath Testing (GBT) is commonly used for diagnosing Small Intestinal Bacterial Overgrowth (SIBO). However, patients present with clinical symptoms (sx) and their predictive value for a positive (pos) GBT remains unclear. AIM: To assess the prevalence of self-reported GI sx profiles in adults with unexplained upper GI symptoms and correlate this with GBT. Methods: We retrospectively analyzed all adult patients presenting to a single medical center from 6/20 –12/21 for GBT. A pos GBT for SIBO was defined by the North American Consensus (≥20 ppm H2 increase over basal within 90 minutes (mins) and/or ≥10 ppm CH4 at any time). Pts completed a GI questionnaire assessing 17 key sxs: regurgitation (RG), chest pain (CP), heartburn (HB), belch (BH), abdominal pain (AP), bloating (BT), gas (GS), nausea (NA), vomit (VT), abdominal cramps (AC), indigestion (ID), distension (DN), fullness (FL), early satiety (ES), diarrhea (DR), constipation (CN), and fatigue (FT). Sxs measured by frequency (None, < 1/week (wk), 1/wk, ≥ 1/wk), intensity (None, Mild, Moderate, Severe), and duration (None, < 10 mins, 10 – 30 mins, ≥ 30 mins). Sub-groups, GERD, Dyspeptic, and IBS were constructed assessing the mean summation scores for Frequency, Intensity, and Duration between negative (neg) and pos GBTs. Univariate analysis and logistic regression analysis was performed. A P-value of < 0.05 was considered statistically significant. Results: We evaluated 184 pts, mean age of 51.7 (18 – 85), 78.3% female, 82.8% Caucasian, and mean BMI of 27.2. No significant differences were seen between GI sx frequency and intensity categories for pts with GBT neg and pos groups. A trend for a higher percentage of pos GBT pts reporting FL (P=0.09) and DR (P=0.07) in sx duration categories of 10 -30 mins (FL: 30.8% vs 27.5%; DR: 31.8% vs 23.4%) and ≥ 30 mins (FL: 51.3% vs 41.3%; DR: 37.2% vs 22.1%) was noted. Regression analysis showed a trend toward a pos GBT for VT frequency of Less than 1/wk: (P = 0.07; OR = 2.99; 95% CI: 0.91, 9.83) and NA intensity of Severe: (P = 0.09; OR = 2.24; 95% CI: 0.86, 5.84). No differences were seen between the GERD, Dyspeptic, and IBS sub-groups (Figure). Conclusion: The pre-test predictability of GI symptoms as it relates to a pos or neg GBT is poor. The yield for a pos GBT varies between 31%-55% and depends on age. In routine practice, when the index of clinical suspicion is high for SIBO, clinicians should consider objective testing with GBT rather than empiric antibiotic treatment.Figure 1.: (Left) Mean Summation Scores for Self-Reported Frequency, Intensity, and Duration Responses to GI Symptoms Relative to the Gastroesophageal Reflux Disease (GERD), Dyspeptic, and irritable bowel syndrome (IBS) Groups. (Right) Prevalence of Positive Glucose Breath Test (GBT) Among Age Cohorts and Glucose Breath Test Measurements

Artificial intelligence and machine learning (AI/ML) is becoming increasingly more accessible to biomedical researchers with significant potential to transform biomedicine through optimization of highly-accurate predictive models and enabling better understanding of disease biology. Automated machine learning (AutoML) in particular is positioned to democratize artificial intelligence (AI) by reducing the amount of human input and ML expertise needed. However, successful translation of AI/ML in biomedicine requires moving beyond optimizing only for prediction accuracy and towards establishing reproducible clinical and biological inferences. This is especially challenging for clinical studies on rare disorders where the smaller patient cohorts and corresponding sample size is an obstacle for reproducible modeling results. Here, we present a model-agnostic framework to reinforce AutoML using strategies and tools of explainable and reproducible AI, including novel metrics to assess model reproducibility. The framework enables clinicians to interpret AutoML-generated models for clinical and biological verifiability and consequently integrate domain expertise during model development. We applied the framework towards spinal cord injury prognostication to optimize the intraoperative hemodynamic range during injury-related surgery and additionally identified a strong detrimental relationship between intraoperative hypertension and patient outcome. Furthermore, our analysis captured how evolving clinical practices such as faster time-to-surgery and blood pressure management affect clinical model development. Altogether, we illustrate how expert-augmented AutoML improves inferential reproducibility for biomedical discovery and can ultimately build trust in AI processes towards effective clinical integration.

Abstract Due to the intricacies surrounding taste profiles, one's view of good wine is subjective. Therefore, it is advantageous to provide a more objective, data‐driven way to assess wine preferences. Motivated by a previous study that modeled wine preferences using machine learning algorithms, this work presents an ensemble approach to predict a wine sample's quality level given its physiochemical properties. Results show the proposed framework out‐performs many sophisticated models including the one recommended by the motivational study. Moreover, the proposed framework offers a simple variable importance strategy to gain insight as to the relevance of the predictor variables and is applied to both simulated and real data. Given the predictive power of using ensembles, especially when they can be interpretable, practitioners can use the following approach to provide an accurate and inferential perspective towards demystifying wine preferences.

Changes since last release: Fix VGG imagenet download. Add numpy as dependency. Convert generators to Keras <code>Sequence</code>s. Float16 support. Expose learning rate parameter. Add validation loss as optional step.

Currently, a straightforward fabrication technique for the development of soft actuators to explore their potential in robotic applications using environmentally compatible raw materials represents an important challenge. A conventional conducting polymer, such as polypyrrole (PPy), shows promising conductivity for such applications. This study presents the synthesis of PPy/polyvinyl alcohol (PPy/PVA)-based ion exchange polymer films containing PEDOT:PSS/SWNT/IL electrodes that undergo conformational changes in response to the applied voltage. Four types of ionic polymer actuator films with different sizes of PPy nanoparticles were fabricated to investigate the size-dependent electromechanical actuation performance. The aim of this study is to design and develop a stable, flexible, and reliable film actuator for robotic applications. Scanning electron microscopy and transmittance electron microscopy were performed to observe the surface morphology and detailed structure of the fabricated actuator films. The current density and ionic conductivity are demonstrated by the cyclic voltammogram and linear sweep voltammogram, respectively. The enhanced values of the water uptake, ion exchange capacity and ionic conductivity of the PPy/PVA polymer composite films enhanced the electrical properties and the tip deflection performance as compared to those of the other reported expensive perfluorinated polymer-based membrane actuators. A two finger-based micro gripping device was also developed, in which both the fingers were made up of the O-PPy/PVA/EL-based ion exchange polymer films. This mechanically stable and flexible film actuator fabricated via a synergistic combination of PPy/PVA composition containing PEDOT:PSS/SWNT/IL electrode surfaces possesses a substantial potential as an actuator material for micro robotic applications.