93887 research outputs found
Sort by
LeTO: Learning Constrained Visuomotor Policy With Differentiable Trajectory Optimization
Get PDFThis paper introduces LeTO, a method for learning constrained visuomotor policy with differentiable trajectory optimization. Our approach integrates a differentiable optimization layer into the neural network. By formulating the optimization layer as a trajectory optimization problem, we enable the model to end-to-end generate actions in a safe and constraint-controlled fashion without extra modules. Our method allows for the introduction of constraint information during the training process, thereby balancing the training objectives of satisfying constraints, smoothing the trajectories, and minimizing errors with demonstrations. This “gray box” method marries optimization-based safety and interpretability with powerful representational abilities of neural networks. We quantitatively evaluate LeTO in simulation and in the real robot. The results demonstrate that LeTO performs well in both simulated and real-world tasks. In addition, it is capable of generating trajectories that are less uncertain, higher quality, and smoother compared to existing imitation learning methods. Therefore, it is shown that LeTO provides a practical example of how to achieve the integration of neural networks with trajectory optimization. We release our code at https://github.com/ZhengtongXu/LeTO. Note to Practitioners—LeTO is driven by the goal of developing an imitation learning algorithm capable of generating safe and constraint-satisfying robotic behaviors. The idea of imitation learning is to enable the robot to learn from human demonstrations of certain tasks. Subsequently, the robot is able to autonomously perform the learned tasks on its own. Thanks to the powerful representational and fitting capabilities of neural networks, imitation learning can let robots perform complex manipulation tasks. However, neural networks often exhibit a certain level of uncertainty and lack theoretical safety guarantees. For robotic systems, it is crucial that robot behaviors meet specific constraints; otherwise, the system may not be sufficiently reliable. Therefore, we introduce LeTO, an approach that integrates trajectory optimization with neural networks to generate actions that not only achieve manipulation tasks, but also comply with constraints. This improves the interpretability, safety, and reliability of robot policies acquired through imitation learning, facilitating their deployment in scenarios with high safety requirements
Course-based Undergraduate Research Experiences (CURE) in Engineering Technology
Get PDFAlthough most universities provide excellent research experiences for outstanding undergraduate students, the number of interested students typically outstrips the supply of faculty and graduate student mentors. Addressing this shortfall is the reason Course-based Undergraduate Research Experiences (CUREs) were created. CUREs face obvious challenges because they operate at a larger scale than the traditional apprentice-based model for undergraduate research, creating resource issues for experimental research that requires equipment, laboratory space, and staff oversight. To help address these hurdles, one university has created a professional development program that provides training, collegial mentoring, and financial support to interested faculty. This paper provides an overview of the CURE program from the perspective of a faculty member who has created CURE content in Engineering Technology and served as a mentor for other faculty pursuing this style of teaching. The purpose is to stimulate discussions of best practices and encourage new faculty to participate
Lifetimes of stable solutions to the planar three-body problem, revealed in patterns of fractal art
Get PDFThis preliminary study explores statistical prediction of the lifetimes of apparently stable orbits of three particle masses, mutually attracted by gravity in the planar three body problem. Most orbits eventually break up, commonly in an end state with two of the bodies orbiting each other and moving together in one direction, with the third body moving in the opposite direction. Studies of such escape events led to a novel hypothesis, emerging from the intersection of probability, statistics, chaos, and fractal art. Statistical estimates of a constant probability of breakup per unit time allow prediction of the times of breakup of three body systems and can be applied to the open question of whether there are some three-body systems whose lifetimes are infinite. Initial conditions for the planar three body problem can be represented in a multi-dimensional feature space. If all three bodies have zero velocity at time zero and the initial positions of two of the bodies remain constant, variable initial positions of the third body in the plane can be defined in a two-dimensional subspace. In numerical simulations, classical Newtonian accelerations, speeds, and positions of all three bodies are calculated as functions of time. Maps of the plane of initial conditions indicating stable systems (in black) vs. broken systems with escape of one or more bodies unstable (in white) resemble fractals. Presented are a framework and sample calculations to test the hypothesis that there is a constant local probability of breakup per unit of time. This probability is estimated after a fixed simulation duration by sampling the measured proportion of points with stable orbits in several arbitrarily small windows surrounding a particular indexed point in the fractal plane. The measured proportions of stable systems, f(n), vary as a function of the number of time steps, n , such that for a range of window sizes, their mean value f̅(n) (1−p)n for a small, local probability, p , of breakup per unit of time. Estimated values of p remain nearly constant over many orders of magnitude of sample window size and over a range of simulation durations, n . This constant probability predicts a gradually falling values of the surviving proportions of such systems, f(n), which approach zero as n approaches infinity
Explainable artificial intelligence to interpret spatially-explicit impacts of future climate change on species distribution
Get PDFBiodiversity is essential for maintaining ecosystem balance and functionality, providing vital services such as climate regulation. The rapid decline in biodiversity, driven by habitat loss, habitat fragmentation, and climate change, poses significant threats to ecosystems. Climate change, in particular, is fundamentally altering habitats, leading to shifts in species distributions. However, existing research often lacks decomposed contribution analyses, particularly spatially, for a changing individual environmental attributes when modeling species distribution as an aggregate result of all factors and their interactions. Such analyses are crucial for identifying climate refugia and prioritizing conservation efforts. Taking endangered mammal species as an example, this study addresses this gap by employing species distribution modeling (SDM) and the post-hoc interpretability method, Shapley values, to analyze how future environmental variables are likely to reshape habitat suitability spatially. Our findings indicate that by 2070, some regions in North America, Europe, and Australia will become suitable for many species due to changes in annual mean temperature, while extensive areas in the Amazon and Congo rainforests will become less suitable. Annual mean precipitation is also projected to drive worsening conditions for local species, particularly in South America and central Africa. Our analysis demonstrates the effectiveness of explainable AI (xAI) techniques, such as Shapley values, in elucidating the future impacts of climate change by accounting for the interactions between environmental attributes. We identify a spatial analysis tool to develop conservation strategies targeted at the environmental attribute level, aimed at mitigating the diverse impacts of climate change on global biodiversity
Updating Cost Allocation and Revenue Attribution
Get PDFThis report presents a synopsis of the results from the 2024 study commissioned by the Indiana Department of Transportation (INDOT) in fall 2023 at the request of the Indiana General Assembly.
Expenditures: Lighter vehicle classes saw their share of cost responsibility decline between 2015 and 2024, while the heavier vehicle classes saw their responsibilities increase within this period. This change is explained by a diametric shift in expenditure type patterns between the two eras from construction-dominant to maintenance-dominant expenditures. In this context, it is worth noting that these expenditure types have different ratios of attributable costs to common costs.
Revenues: Fifty-two percent (52%) of all user and non-user revenues are generated at the state level, 36% at the federal level, and 13% at the local level. Vehicle Classes 2 and 9 still contribute the highest shares of revenues—42% and 22%, respectively. Vehicle Class 3 contributes 21% of the revenues, while all other vehicles contribute less than 10% each. Vehicle Class 13 contributes the lowest percentage share at 0.1%. Across the two eras (2015 vs. 2024), Class 2 vehicles saw their revenue share decline from 47% to 42%, but Vehicle Class 9 increased from 20% in the earlier study (2015) to 22% in the current study (2024). Vehicle Class 3 held steady at 21% in both periods, while Classes 5 and 6 saw marginal increases.
Equity Ratios: The equity ratio results follow a trend that is like those of past studies in Indiana and elsewhere. Generally, the lower vehicle classes are overpaying their share of cost responsibility and the higher vehicle classes are underpaying their share of cost responsibility. Notable shifts in equity ratios between the previous-era study and the current-era study were observed. Several lighter vehicle classes increased their equity ratios, while the heavier vehicles saw their equity ratios decline significantly between the two periods. EVs in Class 2 and Class 3 generally have lower equity ratios than their ICEV counterparts, a finding that is intuitive because EVs are associated with relatively higher damage but slightly lower, or similar, revenue contributions. For the forecast years (2030 and 2035), it was observed that EVs in Vehicle Classes 2 and 3 will be underpaying their share of the cost responsibility, while those in Classes 4 and 9 will be overpaying. The current EV fee that exists for these vehicles (if left unchanged) will not adequately cover their cost responsibility or recover the lost fuel tax revenues in 2030 and 2035
Cross-sectional study of personal protective equipment use, training and biosafety preparedness among healthcare workers during the first months of the SARS-CoV-2 pandemic in Brazil
Get PDFObjectives Brazil has high rates of COVID-19 and tuberculosis among healthcare workers (HCWs). Personal protective equipment (PPE) is essential for their protection. We aimed to evaluate PPE use, training, and preparedness among HCWs in the early months of the SARS-CoV-2 pandemic in Brazil.
Methods A cross-sectional study was performed using questionnaires available to HCWs through a website created to provide PPE guidelines. χ2 test and robust Poisson regression identified factors associated with HCWs treating COVID-19 patients (TCOVID-19), lack of training on PPE use and N95 respirator reuse. The speech content of open-ended questions was analysed.
Results We analysed 1410 questionnaires collected from April to July 2020 representing 526 Brazilian cities. HCWs-TCOVID-19 had fewer years of work experience, were more likely to reuse PPE, and reported higher stress levels and lower biosafety at the workplace than HCWs not TCOVID-19 patients. Fearful concerns, limited PPE access and pandemic unpreparedness were common among HCWs. Lack of PPE training was associated with the profession and no N95 respirator fit tests. N95 reuse during the pandemic, common to 78% of the HCWs, was associated with the reuse of PPE during the pandemic and reuse of N95 before the pandemic.
Conclusions We report the unpreparedness of HCWs and institutions to handle the pandemic, with low rates of training and N95 respirator fit testing and high PPE reuse. N95 reuse was a pre-established practice. This chronic unpreparedness to deal with airborne pathogens may have contributed to one of the highest global rates of tuberculosis and COVID-19 among HCWs
Protocol for recording neural activity evoked by electrical stimulation in mice using two-photon calcium imaging
Get PDFElectrical stimulation provides a clinically viable approach for treating neurological disorders. Here, we present a protocol for recording neural activity evoked by electrical stimulation in mice using two-photon calcium imaging. We detail steps for chronically implanting a head fixation bar, a stimulating electrode, and a glass imaging window.We additionally describe the procedures for viral injections and awake head-fixed recordings
Stretching the Limits of MRI—Stretchable and Modular Coil Array Using Conductive Thread Technology
Get PDFObjective: We propose a modular stretchable coil design using conductive threads and commercially available embroidery machines. The coil design increases customizability of coil arrays for individual patients and each body part. Methods: Eight rectangular coils were constructed with custom-fabricated stretchable tinsel copper threads incorporated onto textile. Tune, match, and detune circuits were incorporated on the coil. A hook-and-loop mechanism was used to attach and decouple the modular coils. Phantom and in vivo scans at various anatomical flexion angles were acquired to highlight performance, and a temperature test was performed to verify safety. Results: In vivo MRI experiments demonstrate high sensitivity and coverage of each anatomy. As the coils are stretched, the sensitive volume increases at a rate of 10.93 mL/cm2. The SNR reduction of a single coil was greater during compression than when stretched, but this did not affect image quality for the array. The modularity of the array allows for adaptability for any anatomy with simple on-demand adjustment to the number and position of coil elements. Conclusion: The images demonstrated high sensitivity and coverage of the stretchable array for various anatomies and flexion angles. Stretching the coils increases the sensitive volume, allowing for a larger region to be effectively imaged. The resonance shift and SNR decrease during stretch and compression support further investigation of methods to reduce frequency shift in stretchable coils. Significance: The proposed array design allows for highly stretchable, flexible, modular, and conformal patient-centered coils that allow for increased imaging quality, greater comfort, and rapid production
Evaluating an Interdisciplinary and Multi-Pedagogical Approach to Equipping Students to Create Social Change
Get PDFModern social problems are complex, multifaceted, and challenging to solve. Scholars are increasingly applying the concept of social innovation as a path to addressing social issues. Social innovation is an interdisciplinary framework for producing social change that requires creativity, problem-solving skills, and collaboration across systems. Higher education is progressively understanding the need to provide interdisciplinary educational opportunities for students; however, little is known about the effectiveness and impact of providing interdisciplinary learning experiences grounded in a social innovation framework. This article describes and analyzes an interdisciplinary summer fellowship program focused on social innovation for graduate students in social work, business, and the humanities and social sciences. The program employed multiple pedagogical approaches, including classroom-based instruction, field learning, and interdisciplinary teamwork. We used qualitative and quantitative pre- and post-evaluation student feedback to examine students’ learning and overall experiences. We found that the fellowship was a dynamic learning experience, through which students strengthened their communication skills and translated academic concepts into practical ideas. The experience also impacted the students’ career trajectories, influencing students to pursue careers that involved working toward social progress in a variety of ways