Explore the vast range of titles available.

Assessing the resilience of road networks in high‐intensity urban development areas is crucial for ensuring sustainable urban growth in the face of increasing traffic demands. Underground road networks are a key solution to alleviating surface traffic congestion and optimizing urban spatial utilization. By enhancing shared mobility on the surface, these networks contribute to reducing vehicle emissions and mitigating environmental pollution. This study explores the optimal conditions for initiating underground road network construction in high‐density development areas. Using a spatiotemporal consumption model, the research calculates the maximum traffic capacity of the network while considering land use characteristics to assess traffic demand. The resilience of these road networks is evaluated through structural indicators that measure their resistance to damage and overall stability. The findings indicate that when the demand‐to‐capacity ratio of the road network ranges from 0.865 to 0.870, the existing road capacity becomes inadequate, necessitating the construction of underground networks to alleviate surface congestion. This study provides both theoretical guidance and practical insights for the planning and development of underground road networks, along with strategies to improve surface environmental quality.

Aims To assess the extent to which biomedical outcomes and cardiovascular risk profile were improved in the management of Chinese patients with type 2 diabetes enrolled in the metabolic management center (MMC) program. Materials and Methods We performed propensity score matching of diabetic patients in the MMC program for at least 12 months to those with diabetes under usual primary care, based on age, sex, fasting plasma glucose (FPG) level, and diabetes duration. Difference‐in‐difference analysis was conducted to compare changes in biomedical outcomes, attainment of treatment targets, and cardiovascular disease (CVD) risk reduction. Results Of 557 pairs of diabetic patients matched 1:1 (n = 1,114), the MMC cohort exhibited greater improvements in FPG (−0.84 mmol/L, 95% confidence interval [CI] −1.22 to −0.46, P < 0.001), diastolic blood pressure [BP] (−2.08 mmHg, 95%CI −3.21 to −0.94, P < 0.001), body mass index [BMI] (−0.29 kg/m2, 95%CI −0.51 to −0.07, P = 0.009), low‐density lipoprotein cholesterol (0.13 mmol/L, 95%CI 0.04–0.23, P = 0.008), high‐density lipoprotein cholesterol (0.05 mmol/L, 95%CI 0.01–0.08, P = 0.017), and 10‐year CVD risk (Framingham CVD risk, −0.94%, 95%CI −1.71 to −0.17, P = 0.017; atherosclerotic CVD risk, −0.77%, 95%CI −1.34 to −0.20, P = 0.009) when compared to the usual primary care cohort after adjustment for confounders. More patients in the MMC cohort achieved treatment targets with lifestyle modifications than their counterparts under primary care. Conclusions Enrolment in the MMC program appears promising in the management of FPG, BP, BMI, lifestyle, and CVD risk in diabetic patients, suggesting the necessity of incorporating the MMC program into routine primary care. We assessed the extent to which biomedical outcomes and cardiovascular risks were improved in the metabolic management center (MMC) program compared to usual primary care. Enrolment in the MMC program appears promising in the management of fasting plasma glucose, blood pressure, body mass index, lifestyle, and cardiovascular disease risk in type 2 diabetic patients.

The built environment is among the critical factors in older adults’ travel behavior, and a favorable built environment can encourage them to travel and engage in various activities. Existing studies have mostly focused on exploring the correlation between the built environment and travel behavior, ignoring the heterogeneity between the two at different times of the day. In this study, we conducted structured, face-to-face interviews in the Nanjing (China) metropolitan area to investigate the time consumed per trip by older adults using various travel modes and used the structural equation and random forest models to explore the relationship between the built environment and older adults’ travel time. The results demonstrated that older adults had different perspectives on travel during different time periods. Different environments and the convenience of destinations affected their overall satisfaction during travel. We found a nonlinear relationship between the built environment and travel time. Metropolitan street connectivity initially had a positive effect on travel time until a certain threshold or peak, whereafter a gradual decline ensued. This nonlinear relationship also existed between the proportion of green space and the distance to subway stations. These results can guide the retrofitting and construction of age-friendly metropolitan infrastructure facilities that promote older adults’ mobility.

Visceral adipose tissue plays an active role in the pathogenesis of type 2 diabetes and vascular dysfunction. The lipid accumulation product (LAP), visceral adiposity index (VAI), and Chinese VAI (CVAI) have been proposed as simple and validated surrogate indices for measuring visceral adipose tissue. However, the evidence from prospective studies on the associations between these novel indices of visceral obesity and diabetic retinopathy (DR) remains scant. This study aimed to investigate the longitudinal associations of LAP, VAI, and CVAI with incident DR in Chinese patients with diabetes. This was a prospective cohort study conducted in Guangzhou in southern China. We collected baseline data between November 2017 and July 2020, while on-site follow-up visits were conducted annually until January 2022. The study participants consisted of 1403 patients with a clinical diagnosis of diabetes, referred from primary care, who were free of DR at baseline. The LAP, VAI, and CVAI levels were calculated by sex-specific equations based on anthropometric and biochemical parameters. DR was assessed using 7-field color stereoscopic fundus photographs and graded according to the modified Airlie House Classification scheme. Time-dependent Cox proportional hazard models were constructed to estimate the hazard ratios with 95% CIs. Restricted cubic spline curves were fitted to examine the dose-response relationship between the 3 indices of visceral obesity and new-onset DR. Subgroup analyses were performed to investigate the potential effect modifiers. The mean age of study participants was 64.5 (SD 7.6) years, and over half (816/1403, 58.2%) were female. During a median follow-up of 2.13 years, 406 DR events were observed. A 1-SD increment in LAP, VAI, or CVAI was consistently associated with increased risk for new-onset DR, with a multivariable‑adjusted hazard ratio of 1.24 (95% CI 1.09-1.41; P=.001), 1.22 (95% CI 1.09-1.36; P<.001), and 1.48 (95% CI 1.19-1.85; P=.001), respectively. Similar patterns were observed across tertiles in LAP (P for trend=.001), VAI (P for trend<.001), and CVAI (P for trend=.009). Patients in the highest tertile of LAP, VAI, and CVAI had an 84%, 86%, and 82% higher hazard of DR, respectively, compared to those in the lowest tertile. A nonlinear dose-response relationship with incident DR was noted for LAP and VAI (both P for nonlinearity<.05), but not for CVAI (P for nonlinearity=.51). We did not detect the presence of effect modification by age, sex, duration of diabetes, BMI, or comorbidity (all P for interaction>.10). Visceral obesity, as measured by LAP, VAI, or CVAI, is independently associated with increased risk for new-onset DR in Chinese patients with diabetes. Our findings may suggest the necessity of incorporating regular monitoring of visceral obesity indices into routine clinical practice to enhance population-based prevention for DR.

Background: Diabetic macrovascular and microvascular complications often coexist and may share similar risk factors and pathological pathways. We aimed to investigate whether 10-year atherosclerotic cardiovascular disease (ASCVD) risk, which is commonly assessed in diabetes management, can predict incident diabetic nephropathy (DN) and retinopathy (DR) in patients with type 2 diabetes mellitus (T2DM).Methods: This prospective cohort study enrolled 2,891 patients with clinically diagnosed T2DM who were free of ASCVD, nephropathy, or retinopathy at baseline in the Guangzhou (2017–2022) and Shaoguan (2019–2021) Diabetic Eye Study in southern China. The 10-year ASCVD risk was calculated by the Prediction for ASCVD Risk in China (China-PAR) equations. Multivariable- adjusted Cox proportional hazard models were developed to estimate hazard ratios (HRs) with 95% confidence intervals (CIs). The area under the receiver operating characteristic curve (AUC) was used to evaluate predictive capability.Results: During follow-up, a total of 171 cases of DN and 532 cases of DR were documented. Each 1% increment in 10-year ASCVD risk was associated with increased risk of DN (pooled HR, 1.122; 95% CI, 1.094 to 1.150) but not DR (pooled HR, 0.996; 95% CI, 0.979 to 1.013). The model demonstrated acceptable performance in predicting new-onset DN (pooled AUC, 0.670; 95% CI, 0.628 to 0.715). These results were consistent across cohorts and subgroups, with the association appearing to be more pronounced in women.Conclusion: Ten-year ASCVD risk predicts incident DN but not DR in our study population with T2DM. Regular monitoring of ASCVD risk in routine diabetes practice may add to the ability to enhance population-based prevention for both macrovascular and microvascular diseases, particularly among women.

Despite advancements in robot learning demonstrating diverse capabilities across various hardware platforms, many systems remain limited to pick-and-place style tasks, characterized by low precision and high compliance. These tasks typically do not require the policy to reason about contact between the object and its environment, which limits the generalizability of such methods to more complex settings where contact reasoning is essential. This limitation stems from two main factors: (1) kinematics – many systems use manipulators limited to simple grasping motions and hence can only obtain action data of simple interactions; and (2) sensing – their hardware platforms often lack contact sensing capabilities and fail to provide enough information in the observations. Furthermore, policies learned by these systems often exhibit low success rates, limiting their practical utility in real-world deployments. A common challenge is distribution shift—a mismatch between the training data and deployment conditions—which undermines the performance of data-driven methods in uncontrolled environments. To address these challenges, this dissertation investigates several key aspects critical to enabling more effective data collection for contact-rich manipulation tasks, ultimately improving policy deployment performance. We begin by introducing novel policy learning pipelines to learn an extremely difficult task--delicate object in-hand rotation. Our proposed learning pipeline starts with an imitation learning step followed by an off-policy on-robot RL fine-tuning using semi-sparse rewards. To achieve this, we use a versatile dexterous robotic hand, ROAMHand3, which is equipped with a multimodal tactile sensor, SpikeATac. This shows the potential of dexterous manipulation using robots that are carefully designed by humans. Next, we explore how to directly optimize hardware configurations for task performance using reinforcement learning, enabling co-design of morphology and control. Finally, we present a framework for leveraging human assistance efficiently during policy deployment, demonstrating how minimal but strategic human interventions can significantly enhance real-world success rates. Through these proposed methods, we argue that data for contact-rich manipulation tasks can be improved along multiple dimensions and they are as important as aspects (e.g., scene, object and task diversity) that other works emphasize. Enhanced sensing capabilities enable more robust performance under environmental perturbations by providing richer and more reliable feedback. Task-optimized kinematics -- achieved through joint design and control optimization -- expand the robot’s reachable workspace and facilitate different interaction modes with diverse objects. Finally, human-in-the-loop policies provide targeted corrections in failure-prone states, enabling the system to recover from suboptimal behaviors and adapt more effectively during deployment.

We introduce the Universal Manipulation Policy Network (UMPNet) -- a single image-based policy network that infers closed-loop action sequences for manipulating arbitrary articulated objects. To infer a wide range of action trajectories, the policy supports 6DoF action representation and varying trajectory length. To handle a diverse set of objects, the policy learns from objects with different articulation structures and generalizes to unseen objects or categories. The policy is trained with self-guided exploration without any human demonstrations, scripted policy, or pre-defined goal conditions. To support effective multi-step interaction, we introduce a novel Arrow-of-Time action attribute that indicates whether an action will change the object state back to the past or forward into the future. With the Arrow-of-Time inference at each interaction step, the learned policy is able to select actions that consistently lead towards or away from a given state, thereby, enabling both effective state exploration and goal-conditioned manipulation. Video is available at https://youtu.be/KqlvcL9RqKM

Human-in-the-loop (HitL) robot deployment has gained significant attention in both academia and industry as a semi-autonomous paradigm that enables human operators to intervene and adjust robot behaviors at deployment time, improving success rates. However, continuous human monitoring and intervention can be highly labor-intensive and impractical when deploying a large number of robots. To address this limitation, we propose a method that allows diffusion policies to actively seek human assistance only when necessary, reducing reliance on constant human oversight. To achieve this, we leverage the generative process of diffusion policies to compute an uncertainty-based metric based on which the autonomous agent can decide to request operator assistance at deployment time, without requiring any operator interaction during training. Additionally, we show that the same method can be used for efficient data collection for fine-tuning diffusion policies in order to improve their autonomous performance. Experimental results from simulated and real-world environments demonstrate that our approach enhances policy performance during deployment for a variety of scenarios.

Controlling robotic manipulators with high-dimensional action spaces for dexterous tasks is a challenging problem. Inspired by human manipulation, researchers have studied generating and using postural synergies for robot hands to accomplish manipulation tasks, leveraging the lower dimensional nature of synergistic action spaces. However, many of these works require pre-collected data from an existing controller in order to derive such a subspace by means of dimensionality reduction. In this paper, we present a framework that simultaneously discovers a synergy space and a multi-task policy that operates on this low-dimensional action space to accomplish diverse manipulation tasks. We demonstrate that our end-to-end method is able to perform multiple tasks using few synergies, and outperforms sequential methods that apply dimensionality reduction to independently collected data. We also show that deriving synergies using multiple tasks can lead to a subspace that enables robots to efficiently learn new manipulation tasks and interactions with new objects.

Underactuated manipulators reduce the number of bulky motors, thereby enabling compact and mechanically robust designs. However, fewer actuators than joints means that the manipulator can only access a specific manifold within the joint space, which is particular to a given hardware configuration and can be low-dimensional and/or discontinuous. Determining an appropriate set of hardware parameters for this class of mechanisms, therefore, is difficult - even for traditional task-based co-optimization methods. In this paper, our goal is to implement a task-based design and policy co-optimization method for underactuated, tendon-driven manipulators. We first formulate a general model for an underactuated, tendon-driven transmission. We then use this model to co-optimize a three-link, two-actuator kinematic chain using reinforcement learning. We demonstrate that our optimized tendon transmission and control policy can be transferred reliably to physical hardware with real-world reaching experiments.