13979 research outputs found
Sort by
Menschbezogene Auswirkungen der Autonomie auf Planungs- und Handlungsebene in der Mensch-Roboter-Kooperation
Get PDFDie vorliegende Dissertationsschrift befasst sich mit den Auswirkungen von Autonomie von Arbeitspersonen in der Mensch-Roboter-Kooperation in Bezug auf deren Motivation, Leistung, Beanspruchung und die Bewertung der Zusammenarbeit mit dem Roboter. Die Autonomie der Arbeitspersonen in Zusammenarbeit mit einem Leichtbauroboter wurde hierzu in einer empirischen Studie auf Planungs- und Handlungsebene differenziert untersucht. Auf Basis des Stands der Forschung wurde die Studie in einem Mixed-Design konzipiert. Die 59 Versuchsteilnehmenden erfuhren zwei Stufen der Autonomie auf Handlungsebene bei der Montage von insgesamt 20 Baugruppen in Kooperation mit einem Leichtbauroboter. Zusätzlich wurden die Versuchspersonen gleichmäßig auf zwei Gruppen (mit niedriger und hoher Autonomie auf Planungsebene) aufgeteilt. Während der Bearbeitung der Arbeitsaufträge wurde die Leistung der Versuchspersonen anhand der Gesamtdauer eines Zyklus und die physiologische Beanspruchung anhand der Herzschlagrate gemessen, nach jeweils 8 Bauteilen füllten die Teilnehmenden einen Fragebogen zur Bewertung ihres Autonomieempfindens, der Beanspruchung und der Bewertung der Teamzusammenarbeit aus.
Ergebnis der Studie war, dass sich eine Einbindung der Versuchspersonen auf Planungsebene positiv auf das Autonomieempfinden auswirkt. Die Gestaltung der Autonomie auf Handlungseffekte wies hingegen keine signifikanten Effekte auf. Das Autonomieempfinden steht dabei in einem positiven Zusammenhang mit der intrinsischen Motivation und identifizierten Regulation der Versuchspersonen. In Bezug auf die Leistung konnten keine Effekte auf die normierte Ausführungsdauer gezeigt werden. Allerdings zeigte sich ein positiver Effekt der Prozessauswahl auf die Gesamtdauer. Bei insgesamt geringen Werten der Beanspruchungskenngrößen konnte lediglich ein Einfluss des Autonomieempfindens auf die mentale Beanspruchung festgestellt werden. In Bezug auf die Bewertung der Teamzusammenarbeit wirkten sich insbesondere die wahrgenommene Autonomie und selbstregulierte Komponenten der Motivation positiv auf eine Bewertung der Teamzusammenarbeit aus.
Die Ergebnisse der Studie indizieren somit positive Effekte des Autonomieempfindens auf die Motivation und Bewertung der Zusammenarbeit mit dem Roboter und zeigen, dass sich eine Einbindung der Arbeitsperson in die Prozessgestaltung positiv auf die Gesamtleistung des Systems auswirken kann
Life Cycle Assessment of an emerging overhead line hybrid truck in short-haul pilot operation
Get PDFIn order to investigate the contribution of electromobility to climate change, greenhouse gas (GHG) emissions of electric vehicles need to be studied in a life cycle (LC) perspective using the method of Life Cycle Assessment (LCA). Most LCA studies focus on passenger cars, but a high share of GHG emissions in the mobility sector comes from freight transport. Here we present a LCA of an emerging overhead line hybrid truck (OH truck) based on a unique database of real data from an eHighway field trial in Germany. We develop a LC transport model that includes vehicle, infrastructure and use phase of regional freight transport. The assessment of the present pilot operation is used as a reference for the evaluation of further developments using a scenario approach. As a first example, we apply the LC transport model to a scenario with realistic short-term improvements from the field trial. The comparison to freight transport with conventional trucks shows GHG savings of about 22%. From the detailed LC transport model outcomes, we derive the contributions of the vehicle and infrastructure components to GHG emissions as well as further environmental impact categories and we evaluate the operational phase, e.g. by identifying the break-even point for GHG savings in dependence of the utilization frequency. The findings can be used to identify further improvements of the OH truck technology. At the same time, our model enables a thorough investigation of future scenarios, allowing a robust comparison of possible alternatives for decarbonizing freight transport
Three Essays in Energy and Environmental Economics
Get PDFThis dissertation consists of three studies, each examining a different topic in the field of energy and environmental economics. The topics comprise: (i) estimating potentials for greenhouse gas emission reductions of economic sectors, (ii) forecasting the oil production of a region based on historical data from discoveries, and (iii) examining the determinants of electricity price fluctuations. Each of the studies uses a particular statistical method or mathematical model that is specifically adapted to the research question and the data set under investigation.
The first study is a stochastic nonparametric efficiency analysis in which greenhouse gas emissions are included as bad outputs. For seven economic sectors and sixteen European countries, this study estimates greenhouse gas emission reduction potentials, i.e., the quantity of emissions that could potentially be reduced by improvements in productive efficiency. The standard DEA method is extended by a specific bootstrapping procedure used to implement a bias correction and to compute confidence intervals. The magnitudes of the emission reduction potentials are compared with the emission reduction targets for 2030 from the European Commission. The results show that improvements in productive efficiency are a quantitatively important element, potentially allowing for a substantial reduction of greenhouse gas emissions in the European Union.
The second study presents a stochastic model for forecasting for an oil-producing region the amount of undiscovered oil, the future path of oil discovery and that of oil production. The model combines three submodels: (i) an empirically-founded production model at the level of individual oil fields, (ii) a successive sampling discovery model after Kaufman et al. (1975) for forecasting field sizes, and (iii) a stochastic birth process model for forecasting discovery dates. The model is estimated and evaluated for the oil-producing regions of Norway and the U.S. Gulf of Mexico (the latter further split into shallow- and deep-water parts). The results show that the predictions for oil discovery are somewhat too low compared to the actuals for Norway and for the shallow-water Gulf of Mexico, while for the deep-water Gulf of Mexico the predictions are too high. This is similarly reflected in the predictions for oil production.
The third study is a multivariate wavelet analysis of the German wholesale electricity market, which examines the determinants of electricity price fluctuations using daily time series. The possible determinants are coal prices, gas prices, and the residual load (i.e., electricity consumption minus wind and solar generation). The multivariate wavelet method allows for a detailed examination of the relations between the time series in time-frequency space, while also taking into account the interdependencies among the different time series. The results show that the residual load is the key short-run determinant of electricity prices, while coal and gas prices are the key long-run determinants. Also, this study finds that the co-movement relation among the energy prices is time-varying, which is consistent with the findings of other studies (e.g., Sousa et al. (2014); Aguiar-Conraria et al. (2018))
Modernization and Optimization of MPI Codes
Get PDFMPI has become the de facto standard for distributed memory computing since its inception in 1994.
While the MPI standard has evolved to include new technologies like RDMA,
many applications still rely on the original set of MPI operations. This thesis initially
investigates the current usage of MPI. We note that developers underutilize modern MPI
features, as their implementations often are not optimized. On the other hand, as many
users rely on the ”old” MPI features, the MPI implementation developers have no incentive
to optimize implementations for the new features. As a consequence, there is no incentive
for the MPI users to learn the new features, creating a vicious cycle.
To break this cycle, this thesis explores three main approaches: (1) facilitating correctness
checking tool support, (2) modernizing MPI codes with compiler based approaches, and
(3) exploiting compiler knowledge to further optimize the implementation of modern MPI
features.
In order to facilitate the development and improvement of tools aiding with MPI develop-
ment, this thesis introduces the correctness benchmark MPI-BugBench, as a standardized
benchmark to evaluate the real world applicability of such tools. Further we show that
compiler-based automatic modernization methods can encourage early adoption of new
MPI features with minimal programmer effort, for example partitioned operations. Lastly,
compiler knowledge can be utilized in order to further optimize the performance of MPI
implementations, for example in persistent operations. The use of compiler knowledge, in
particular, enables modernization of existing MPI codes without the need of application
developers to rewrite existing MPI codes
Advances in the Theory of Compact Groups and Pro-Lie Groups in the Last Quarter Century
Get PDFThis article surveys the development of the theory of compact groups and pro-Lie groups, contextualizing the major achievements over 125 years and focusing on some progress in the last quarter century. It begins with developments in the 18th and 19th centuries. Next is from Hilbert’s Fifth Problem in 1900 to its solution in 1952 by Montgomery, Zippin, and Gleason and Yamabe’s important structure theorem on almost connected locally compact groups. This half century included profound contributions by Weyl and Peter, Haar, Pontryagin, van Kampen, Weil, and Iwasawa. The focus in the last quarter century has been structure theory, largely resulting from extending Lie Theory to compact groups and then to pro-Lie groups, which are projective limits of finite-dimensional Lie groups. The category of pro-Lie groups is the smallest complete category containing Lie groups and includes all compact groups, locally compact abelian groups, and connected locally compact groups. Amongst the structure theorems is that each almost connected pro-Lie group G is homeomorphic to RI×C for a suitable set I and some compact subgroup C. Finally, there is a perfect generalization to compact groups G of the age-old natural duality of the group algebra R[G] of a finite group G to its representation algebra R(G,R), via the natural duality of the topological vector space RI to the vector space R(I), for any set I, thus opening a new approach to the Hochschild-Tannaka duality of compact group
Semantic segmentation with deep learning: detection of cracks at the cut edge of glass
Get PDFIn this paper, artificial intelligence (AI) will be applied for the first time in the context of glass processing. The goal is to use an algorithm based on artificial intelligence to detect the fractured edge of a cut glass in order to generate a so-called mask image by AI. In the context of AI, this is a classical problem of semantic segmentation, in which objects (here the cut-edge of the cut glass) are automatically surrounded by the power of AI or detected and drawn. An original image of a cut glass edge is implemented into a deep neural net and processed in such a way that a mask image, i.e. an image of the cut edge, is automatically generated. Currently, this is only possible by manual tracing the cut-edge due to the fact that the crack contour of glass can sometimes only be recognized roughly. After manually marking the crack using an image processing program, the contour is then automatically evaluated further. AI and deep learning may provide the potential to automate the step of manual detection of the cut-edge of cut glass to great extent. In addition to the enormous time savings, the objectivity and reproducibility of detection is an important aspect, which will be addressed in this paper
Learning from Imperfect Human Input in Interactive Machine Learning
Get PDFFrom early childhood, humans collect a lifetime of experiences and are, therefore, a valuable source of knowledge. Learning from human teachers in robotics can enable the transfer of this prior knowledge to the robot by providing input, i.e., in the form of task demonstrations, advice, or feedback on robotic actions. This way, instead of being limited to a pre-defined set of programmed skills, a robot’s skill set can be extended and personalized by a human user. Involving humans in the learning process and adapting the robot’s skill set to the respective wishes and needs of the user can additionally enable learning from non-experts and increase the acceptance of intelligent robots. However, in real application scenarios in everyday life or industry with inexperienced users, this user input is prone to error, as these complex environments exhibit a high degree of uncertainty, and users often have to make decisions based on limited information. To enable learning from humans in these real, unstructured environments, it is crucial to detect and handle erroneous or imperfect human input and develop methods that allow robots to leverage and exploit input from human teachers despite its flawed nature. This work approaches learning from imperfect human input from three different angles. First, we analyze human behavioral data corresponding to human uncertainty with the goal of detecting uncertain user input. Existing studies already point to a correlation between human uncertainty and the correctness of responses. This uncertainty is communicated verbally and non-verbally when interacting with another human. Humans are surprisingly good at assessing this communicated uncertainty and use this information to evaluate the content of conversations. We propose a model that recognizes human uncertainty in a similar way based on multi-modal behavioral data, such as speech, eye and head movements, response time, and facial behavior. In contrast to existing methods in the literature, we here focus on multi-modal behavioral cues related to self-reported human uncertainty in human-robot interactions with the goal of learning a human uncertainty classifier applicable in HRI settings as a potential indicator for incorrect human input. In experiments with human participants, we collected multi-modal behavioral data in decision-making tasks with uncertainty and analyze behavioral differences between human-human interactions and human-robot interactions. Evaluations of our developed multi-modal human uncertainty classifier trained on the collected dataset show that it significantly outperforms third-person annotators in accuracy and F1 score. While humans report feeling less observed when responding to a robot compared to a human, behavioral differences did not significantly affect the performance of our proposed uncertainty classification.
Second, we consider learning from sub-optimal human input in the context of Interactive Reinforcement Learning (IRL). Classical Reinforcement Learning enables intelligent agents such as robots to learn new skills by maximizing rewards through interaction with the environment and performing appropriate actions. Human input in the form of feedback on actions or direct action advice in IRL can accelerate learning by benefiting from prior human knowledge of actions and the environment. Related methods often simulate these inputs and assume their optimality. Moreover, most approaches that account for inaccurate advice compute trust in human action advice independent of a state. In reality, however, a high degree of uncertainty or limited understanding of the task can result in incorrect human input. In particular, human input might be inaccurate only in some states while still being useful in others. Therefore, we present LUNAA, an IRL method that enables learning from partially incorrect human action advice by computing a state-dependent measure of trust in human advice. This allows LUNAA to discard incorrect advice for particular states while still profiting from correct advice in other states. Here, we combine three different indicators for potentially incorrect human action advice. Since human uncertainty is related to answer correctness, as described above, we consider behavioral cues related to human uncertainty as one indicator. We combine this human uncertainty estimate with the consistency of action advice and a reward-based indicator retrospective optimality. The resulting trust measure determines whether the human action advice is accepted. In addition, we propose LUNAA-TIP which extends the proposed LUNAA approach by introducing state-dependent trust measures in the policy in addition to the state-dependent trust in human advice. This allows LUNAA-TIP to distinguish between states with a high policy certainty, e.g., based on a high number of state visits, and states with a low certainty where it might be more beneficial to follow human advice. Moreover, LUNAA-TIP utilizes a real-time implementation of the previously proposed multi-modal human uncertainty classifier based on behavioral cues as an indicator for unreliable human action advice. Evaluations in gridworld environments with simulated advice show that LUNAA and LUNAA-TIP outperform a state-independent baseline without a computation of trust in human advice and the policy. In addition, LUNAA-TIP outperforms LUNAA confirming the benefit of an additional state-dependent trust in the policy. In robotic experiments with advice from human participants, we confirm the usefulness of behavioral cues related to human uncertainty as an indicator for unreliable advice and show that LUNAA-TIP is more robust to incorrect human advice compared to a state-independent computation of trust in the policy.
Third, we look at Learning from Demonstration (LfD), particularly learning robotic skills in the form of Behavior Trees (BTs) from a potentially incomplete small set of human video demonstrations. Behavior Trees are a possible representation of robot capabilities that map a state to an action. They are characterized by a high modality, reactivity, and high interpretability and are, therefore, a well-suited skill representation for LfD approaches. Demonstrating a task is a very natural way of teaching for a human instructor and inspired by how humans can learn by observing and imitating others. In human-robot interactions, this type of learning from human demonstrations is particularly suitable for non-experts as it requires no prior knowledge and is less time-consuming compared to other BT-based LfD approaches that require a step-by-step specification of necessary actions. We, therefore, automatically learn a Behavior Tree with action conditions from a limited number of human video demonstrations. In contrast to existing methods, we automatically compute continuous pre- and post-conditions based on visual features and use these to build a reactive BT. While this teaching approach is very intuitive and requires only a few task demonstrations, learning from this small set of natural and potentially sub-optimal demonstrations poses additional challenges. In a preliminary study, we therefore investigate how non-experts demonstrate tasks and vary the demonstrations. We identify three common failure cases of a BT learned from only a few potentially incomplete demonstrations. The pre- and post-conditions allow the robot to recognize these failure cases during the execution of the BT and to resolve them by interacting with the human user. Here, the robot explicitly asks for additional input and improves or extends the initial BT and learned action conditions accordingly. In contrast to existing methods, in this way failure cases caused by imperfect or incomplete demonstrations can be interactively resolved, and repeating the entire teaching process is not necessary. We evaluate the method in a sorting task with a robot arm and human participants and show that it is possible to learn a reactive BT from just a few video demonstrations and to interactively resolve failure cases during runtime.
In summary, this thesis contributes methods to detect, learn from, and overcome problems arising from imperfect human input. We show that taking into account the sub-optimality of human input enables a more effective exploitation of the potential of human input and improves learning in the context of Lf D and IRL. The contributed methods and robotic experiments with participants with mostly little to no robotic experience provide insights into the nature of human input and challenges but also opportunities when learning from human teachers for future work
Survey of Open Data Concepts Within Fundamental Physics: An Initiative of the PUNCH4NFDI Consortium
No full textPUNCH4NFDI (Particles, Universe, NuClei and Hadrons for the NFDI) aims at developing concepts and tools for the efficient management of digital research products in fundamental physics research. At the heart of the research products are scientific data sets that should be made interoperable and available to a broad scientific community and the public for a sustainable usage (“open data”). The first PUNCH4NFDI “Open Data Workshop” gave the opportunity for an initial survey of existing and planned open data initiatives within the PUNCH science field. The paper addresses the conceptual differences and commonalities of the participating communities presented in the workshop. Existing open data collections were presented and discussed. This is an inquiry into the community’s requirements for a better use of open data and in this context also of “Open Science”
Vacuum insulated glazing (VIG) units under wind load—part 1: global deformation and stresses on the outer glass surfaces
No full textThis paper presents the first part of a study on the effect of wind loads on Vacuum Insulated Glass (VIG) units. The study provides background information on VIG and relevant Standards, explains the numerical modelling process, and discusses the implications of the results in relation to European and North American codes and Standards. The focus of the study is on vertical windows and façade installations in low-rise buildings (< 25 m) commonly found in residential buildings. The mechanical behaviour of VIG was analysed using the Finite Element Method (FEM) with respect to various design parameters such as glass pane size, glass thickness, surface pressure magnitude, and edge boundary conditions. The study analysed global deformation as well as the stresses on the outer glass surfaces. The VIG features such as pillar geometry and contact dynamics, and material non-linear effects, were explicitly modelled. In addition, monolithic glass plates were also modelled, and the FEM results of the monolithic cases were in reasonable agreement with an analytical solution obtained from linear thin plate theory. These results highlight the limit of linear behaviour in monolithic plate bending. The centre-of-pane deflection of the VIG was in good agreement with the FEM and analytical solutions of the equivalent thickness monolithic pane, for sample sizes below 800 × 800 mm. However, for larger glass sizes, a deviation was found, and the VIG exhibited a higher plate stiffness than the equivalent thickness monolithic pane. The simulations also showed that the stresses in the glass panes are highly dependent on the edge of glass boundary condition. Finally, the results demonstrated that with appropriate design choices, the VIG can satisfy the Standards requirements for wind load and glass design in both Europe and North America
Ls(Lq)-estimates of the pressure - The proof of Sohr-von Wahl and its impact on mathematical fluid dynamics
No full textThe article On the regularity of the pressure of weak solutions of Navier–Stokes equations by H. Sohr and W. von Wahl (1986) is one of the most-cited papers of the journal Archiv der Mathematik. Our aim is to describe not only the content of the paper, but especially the novelty of its results on maximal regularity of the Stokes operator considered from a point of view of the eighties and from a modern point of view