At German Aerospace Center (DLR) new spacecrafts are designed by engineers within a special room, called Concurrent Engineering Facility (CEF). Here, the design of the first structural configuration of a spacecraft is an essential task, which is usually done in a non-collaborative manner using a Computer Aided Design (CAD) program guided by one of the engineers. First attempts tried to use virtual reality to shift the design process to a more cooperative approach. Nevertheless, these approaches lacked on intuitiveness and immersiveness. In this paper, we apply augmented reality (AR) technology to overcome those deficiencies. We present a new AR based approach to support concurrent engineering processes and investigate the acceptance of interactive collaboration for a spacecraft design scenario. We connect multiple AR head mounted displays over a network to enable real-time collaboration in a shared environment. Furthermore, we implement an AR application, which fetches visualization data from Virtual Satellite 4, an open source data model for early-stage spacecraft design, and allows manipulation of individual spacecraft parts through hand gestures. To evaluate our case study, we conduct a perceptual study with 11 participants à 2-3 persons in 4 groups. We measure the application’s usability, potential usefulness and ability to resolve interdisciplinary conflicts between requirements, such as difficulties of mutual understanding. Our results show that the application is user-friendly and has a supportive user interface. Furthermore, the collaboration and a natural communication was rated as good. In general, the potential usefulness and acceptance in CEF for such technology is given. However, its relevance for solving interdisciplinary conflicts during the spacecraft configuration process is controversial and requires further research, which is discussed in detail.

DLR is using a data model for their Concurrent Engineering studies. This model is intended to be used for the whole life-cycle of a spacecraft. Here, the system is usually decomposed by so called product structures. So far, the used product structures in the model differ between life-cycle Phase A and B. This leads to reuse problems, requires complicated transformations, and potentially incurs information loss. As a consequence, it is required to harmonize these product structures. This paper defines a reasonable subset of Phase B product structures for the application in CE. They will have an increase in model size as well as user interactions. The increase is quantified in this paper and discussed with qualitative observations of two studies where these new product structures have been used.

Augmented reality techniques can be used to support system modeling and industrial operations at different levels, enabling designers and engineers to augment their real environment with relevant virtual content. In aerospace, these techniques are tightly coupled to digital twins. Together, they can enhance scarce operational resources, facilitating skill transfer and knowledge retention. It is possible to define the term digital twin based on conceptual data models as used in model-based system engineering. In this definition, a conceptual data model is used to accompany a product as an unique evolving system model through the whole lifecycle, starting from virtual abstractions and progressing toward a virtual replication of the real-world entity. Throughout the whole lifecycle, digital twins help to analyze or predict system behavior for improving decision-making and avoiding cost-expensive prototyping. In this chapter, we discuss how digital twin representations can leverage on augmented reality approaches and provide an overview of how model-based system engineering can help to maintain information consistency through the different phases of the product lifecycle. In this context, we address different aspects related to the use of augmented reality approaches for digital twins in aerospace, such as overlay precision, interaction, data visualization, and remote collaboration. Our example applications take different phases of the product lifecycle into account, from creation to operation.

In model-based systems engineering projects, engineers from multiple domains collaborate by establishing a common system model. Multi-level modeling is a technique that can be used to model the development from abstract ideas to concrete implementations. However, current multi-level modeling approaches are not adequate for processes with multiple modeling phases that might have to be rearranged later. In this paper, we introduce multi-phase modeling that utilizes concepts of multi-level modeling by considering a description of the expected phase ordering per domain. Constraints aware of this context can express that certain elements are only valid in specific phases without having to determine a concrete phase ordering for a particular model. This enables using multi-phase modeling in flexible workflows, adapting to changing requirements and the definition of access rules in domain notation. We show feasibility of this multi-phase modeling by applying it to multiple real-life systems engineering projects of the aerospace domain.

Convolutional neural networks (CNNs) have recently been very successful in a variety of computer vision tasks, especially on those linked to recognition. Optical flow estimation has not been among the tasks CNNs succeeded at. In this paper we construct CNNs which are capable of solving the optical flow estimation problem as a supervised learning task. We propose and compare two architectures: a generic architecture and another one including a layer that correlates feature vectors at different image locations. Since existing ground truth data sets are not sufficiently large to train a CNN, we generate a large synthetic Flying Chairs dataset. We show that networks trained on this unrealistic data still generalize very well to existing datasets such as Sintel and KITTI, achieving competitive accuracy at frame rates of 5 to 10 fps.

This paper contributes to the topic of spacecraft interface and data rate management in Concurrent Engineering (CE) sessions. At DLR, CE is used together with a CE process for designing new spacecraft. The software Virtual Satellite supports this process. It provides a shared system model to the engineers to exchange design information. Until today, it supports the structural decomposition of the system and the analysis of design drivers such as the mass or power consumption of the spacecraft. During one of the S2TEP studies for a multi-mission platform it was required to have a closer look to power and data interfaces. This paper discusses the state of the art to this topic and derives a generic approach to it. This approach is customized and finally implemented in Virtual Satellite and directly applied in the S2TEP study.

In this paper, we introduce a system to collect product information from manufacturers and make it available in tools that are used for concurrent design of spacecraft. The planning of a spacecraft needs experts from different disciplines, like propulsion, power, and thermal. Since these different disciplines rely on each other there is a high need for communication between them, which is often realized by a Model-Based Systems Engineering (MBSE) process and corresponding tools. We show by comparison that the product information provided by manufacturers often does not match the information needed by MBSE tools on a syntactic or semantic level. The information from manufacturers is also currently not available in machine-readable formats. Afterwards, we present a prototype of a system that makes product information from manufacturers directly available in MBSE tools, in a machine-readable way.

Concurrent Engineering (CE) and Model Based Systems Engineering (MBSE) have increased the efficiency of spacecraft, and satellite design in particular. Early design of satellites in Concurrent Engineering Centers (CEC) has almost become business as usual. However, such progress has still to be achieved for the design of launchers. Applying the same approaches as used for satellites has not led to the same amount of improvement, yet. To address this, DLR initiated the project Concurrent Launch Vehicle Analysis (CLAVA) to investigate the shortcomings and to improve the efficiency of conceptual launcher design and analysis. From an MBSE point of view, investigations show that concurrent modelling requires new Conceptual Data Models. In contrast to designing satellites, they are focused on a much more physical abstraction rather than a functional one. Regarding simulations, it has become clear that the conceptual design phase of launchers requires far more computationally intense simulations in a sequential order. With this knowledge, it is possible to outline a new process for CE studies allowing for concurrent design phases and sequential simulation phases. For this, an adjusted architecture of tools is required as well. The data model used for satellite studies within DLR's Concurrent Engineering Facility (CEF) does not fit to the requirements of launcher design and has been adapted. Additionally, DLR's aeronautics divisions have already made substantial progress in increasing the efficiency of their simulations. They employ automated simulation workflows using a parametric model for information exchange between integrated tools. This approach has been adopted and integrated. This paper outlines how this approach is combined with CE and MBSE concepts used for satellites and addresses the specific requirements of launcher design. It provides details about the database used during CE sessions, and how its information is transferred into the parametric data model used to run the required simulations. The conceptual data model of this database has been adapted to the physical representation of launchers; these changes will also be discussed. Furthermore, the general idea of the workflow and the design of the parametric model will be presented. The paper concludes by providing an outlook of how DLR intends to continue on this work, and further refine the developed tools and processes into daily CE and CEF application.

Introduction Collaboration is a key factor for successful development of spacecraft. It is a broad term and has many meanings because exchange of information is needed all along the development life cycle. A successful design directly depends on sharing necessary information with others and, on the other hand, comprehending and using relevant information as input for your own work. Modeling and simulation is extensively used in spacecraft design and appears in many different varieties throughout the whole development life cycle. It covers a wide range, from simple orbit simulations based on analytical equations to high-fidelity and multidomain simulations using distributed high-performance computers. In today’s large-scale, complex projects, traditional engineering approaches reach...

The finding that very large networks can be trained efficiently and reliably has led to a paradigm shift in computer vision from engineered solutions to learning formulations. As a result, the research challenge shifts from devising algorithms to creating suitable and abundant training data for supervised learning. How to efficiently create such training data? The dominant data acquisition method in visual recognition is based on web data and manual annotation. Yet, for many computer vision problems, such as stereo or optical flow estimation, this approach is not feasible because humans cannot manually enter a pixel-accurate flow field. In this paper, we promote the use of synthetically generated data for the purpose of training deep networks on such tasks.We suggest multiple ways to generate such data and evaluate the influence of dataset properties on the performance and generalization properties of the resulting networks. We also demonstrate the benefit of learning schedules that use different types of data at selected stages of the training process.

A lot of different parties have to communicate and exchange data with each other during the lifecycle of a spacecraft. One example we are looking into is the usage of information from manufacturers in CE studies during the planning phase of a spacecraft. Currently, information from manufacturers is usually offered as a PDF file that describes the technical features of a component. This information must then be entered manually in an MBSE tool by an engineer. That, and finding a fitting component for the mission requirements in the first place, costs time. It also is error prone due to typing mistakes for example. How about a system where an engineer can place a request and get a list of fitting components from different manufacturers - maybe even directly connected to the MBSE tool? To realize such a system, it is necessary that all manufacturers describe their products in a uniform and comparable way. Electronic Data Sheets (EDS) can be a way to realize such a uniform description - not only for the described use case but for many more. A uniform automatic exchange of information about spacecraft components is also relevant in other (also depends on?) phases of the lifecycle of a spacecraft, for example when testing manufactured components. For testing, a different view on the component is relevant. While during planning mass budgets are calculated and compared, for testing it is relevant to have detailed information about interfaces, protocols, and commands. Not only are different views relevant for different phases of the lifecycle but different categories of components have also different sets of relevant parameters. Frequencies for example are interesting for antennas but rather not for batteries. So, there will be neither one EDS that contains the information for one component in every phase of the lifecycle of a spacecraft nor one EDS format that fits all components categories at the same time. Still, for automated communication it is necessary to develop standards for the description of components - it just might not be one standard but a set of standards with a common vocabulary. To ensure the semantic compatibility between different EDS formats we think that ontologies can help. For example, one ontology can describe the semantics of a mechanical Interface Control Document (ICD) and another ontology can describe the semantics of an electrical ICD. But both ontologies can share common parts, in this example the description of the pins of an interface. In software engineering it became common to develop rather small applications that are linked together instead of a huge one that covers all use cases. We think that a similar approach also makes sense for documents and information exchange - to have small documents but a common language that enables different parties to talk with each other without having to understand each other’s domains fully. This concept also encourages the knowledge sharing and reusing the existing ontology.

The lifecycle of a spacecraft follows a process of phases. There are important goals between these phases such as a preliminary design review (PDR). These goals have to be successfully passed by the whole project team. Usually they are connected to contractual conditions, e.g. after the PDR the design is usually settled and the spacecraft will be built. Such contracts consider agreements with industry and suppliers on part and equipment orders, etc. [1] [2] Changing the design after the PDR may require contractual changes. These changes tend to be expensive and have to be avoided. [3] To avoid the aforementioned issues, model based systems engineering (MBSE) has been introduced in satellite design. It is focusing on data bases that provide a conceptual data model (CDM also known as meta model). The engineers start modeling the system within such data bases usually on a functional level. The system model is then used as central source of knowledge for further processes. [3] Such processes may cover on the fly analysis [4], configuration of simulators [5], new ways of modelling including interactive visualization [6] and emerging trends such as mixed reality. This notion of an MBSE approach has been implemented e.g. in DLR’s data base called Virtual Satellite. Until today it is successfully applied in early spacecraft design. [7] Nowadays, the data base is applied beyond the early phases. Therefore it has to deal with more detailed information and has to cope with yet unknown requirements of tomorrow. Accordingly the CDM has become more complex and offers extension mechanisms. [3] The success of these data bases is partly founded in their configuration control capabilities. In fact, the engineers require not just one model of the spacecraft but several. These models are a master model, derived simulator models, or models for the actual satellites one and two. The different models are needed to reflect differences e.g. a different electrical harness for the simulator, or individual command ids for individual satellites. [3] [5] This is handled by the product structures of the data bases. They are standardized to a certain extend in the activities of European Ground Segment – Common Core (EGS-CC) as well as the European Cooperation for Space Standardization Technical Memorandum (ECSS-E-TM) 10-23. [8] [9] These structures are used for a hierarchical decomposition of the system. The first tree modelled, is usually a product tree. The engineers are using this tree for an initial description of the required parts such as a reaction wheel (RW) or magnetic-torquer (MTQ). They don’t yet model every instance of such a part. Instead, they are modelling them as an idea of a type. The second tree is the configuration tree where these types are instantiated into a virtual configuration of several RWs and MTQs. The final trees are the assembly trees, which are based on the configuration trees but representing how actual satellites are build. They reflect the assembly e.g. of an actual satellite number one and two. Since all trees, including their parts, are based on each other, defining the mass of the RW once in the product tree will update all its further instances in the configuration and assemblies as well. Override functionality allows changing the values when needed or to combine the information with so called realizations. These realizations represent the actual ordered parts that have been delivered. E.g. their calibrations are measured and the best fitting parts are now assigned to the assembly. The information such as a mass is modelled by so called engineering categories. They are based on what is known as type/object pattern. [10] [11] An engineering category defines a property such as a mass and assigning this category to an element in the product tree instantiates it. Now information can be stored in the instance of this property. [3] Even though successful, there are some drawbacks, where Multi-Level Modeling promises some reasonable improvements. For example, an engineering category for storing geometric information of a part consists of a position, a size and a shape. These three properties make sense at configuration level, but not yet at product level. At product level or type level the position is simply not yet known. By today this is handled by either providing arbitrary values or by complex class inheritance hierarchies. Nevertheless such handling is a workaround rather than a proper solution to such problems. Multi-Level Modelling and the idea of potencies and deep instantiation [12] in particular, seems to offer a solution. Assigning a potency of two for the position property creates awareness of this property already at product level. The actual value of that property can now be set starting from configuration level. Considering another example based on an engineering category for tele-commands, it consists of a purpose, e.g. RW turn on, an equipment identifier as well as a satellite identifier. With the concept of potency and deep Published in J.P.A. Almeida, U. Frank and T. Kühne, “Multi-Level Modelling (Dagstuhl Seminar 17492”, Schloss Dagstuhl--Leibniz-Zentrum für Informatik, Dagstuhl Reports 2018, Volume 7, Number 12, pp 24-27, DOI: 10.4230/DagRep.7.12.18, http://drops.dagstuhl.de/opus/volltexte/2018/8675 Potential of Multi Level Modelling in Model Based Systems Engineering A “National Research Lab” Perspective Philipp M. Fischer, DLR Software for Space Systems and interactive Visualization, Lilienthalplatz 7, 38108 Braunschweig Page 2 instantiation, engineers are aware of all three properties already at product level. The actual necessary information needs to be provided at the stages of configuration and assembly. This approach works well with the accepted set of product structures. Nevertheless current work indicates that there might be further trees needed in future applications. Introducing a new integration tree in between configuration and assembly breaks the potency mechanism for the tele-command example. A decrease of that potency with every level of instantiation is not suitable. A fix to this issue could lead in the direction of context aware potencies. At the moment, the MBSE data bases do not yet apply such Multi-Level Modelling. Nevertheless, it can be seen that certain directions of Multi-Level Modelling could improve the overall modeling activities. For sure this view is a highly practical driven adoption of the theories of Multi-Level Modelling and potentially breaks some of the clear cut semantics. Still it shows that these theories provide some answers to the problems of spacecraft related models of today. In general, the described idea of applying potencies and deep instantiation to the concept of engineering categories looks promising. In order to prove its applicability, further research is needed and some first prototypes are envisaged for evaluation.

In this paper we propose an integrated immersive augmented reality solution for a software tool supporting spacecraft design and verification. The spacecraft design process relies on expertise in many domains, such as thermal and structural engineering. The various subsystems of a spacecraft are highly interdependent and have differing requirements and constraints. In this context, interactive visualizations play an important role in making expert knowledge accessible. Recent immersive display technologies offer new ways of presenting and interacting with computer-generated content. Possibilities and challenges for spacecraft configuration employing these technologies are explored and discussed. A user interface design for an application using the Microsoft HoloLens is proposed. To this end, techniques for selecting a spacecraft component and manipulating its position and orientation in 3D space are developed and evaluated. Thus, advantages and limitations of this approach to spacecraft configuration are revealed and discussed.

Spacecraft design and development is a complex Task which requires a long lifecycle and many phases of iterative construction. Thus, there are usually many versions of configuration design evolving over time. A clear understanding and comparison of these different Versions of spacecraft configurations can bring benefits to both stakeholders and domain experts of the project. However, usually the configuration and system data is stored in textual format in tables and in files which is not efficient to compare results by checking all the subsystem and parameters in different versions one by one. As an alternative, this paper gives a detailed insight how to compare versions by means of 3D visualization and Virtual Reality (VR) technologies. The basis of our approach is the Virtual Satellite (VirSat) which is the standard tool for concurrent engineering studies at the German Aerospace Center (DLR). The key performance indicators of an early space mission design are the mass, the power consumption, and the temperature. In order to Highlight differences of these quantities and distribution of key performance indicators, the VirSat uses color coding for more intuitive understanding. To analyze the changes of geometry information, additional two visualization modes have been integrated for comparing and indicating the differences between versions. In the first mode, two versions of configurations are visually overlapped. The differences are indicated by special colors. In the second mode, different versions of the design yield a successive animation in order to follow the design evolution. Both modes require model transformation from VirSat's central system data model to visualization model which is used for appropriate visualization approaches. The paper gives a detailed architectural overview and discusses benefits and future opportunities.

Family firms form the majority of enterprises in almost all national economies. While public corporations consist of the two components ownership and business, family firms have a third component: the family. In the past, brand management and family firms have been seen as two separate fields of research. This empirical study focuses on brand management of family firms, which can be seen as a complex interdisciplinary field of research. Taking the complexity into account, a model based on the brand identity model of Burmann, Halaszovich, and Hemmann (Identitätsbasierte Markenführung. Springer-Gabler, 2012) and qualitative interviews that were conducted with German family firm managers, is developed. This model, the so-called "markencloud", categorizes the different aspects of brand management in family firms and gives family firm owners additional hands-on information on their company's brand. The "markencloud" does not only illustrate complexity from a scientific point of view, it also works as a practitioners' tool to support a family firm's brand management.

Dental radiography plays an important role in clinical diagnosis, treatment and surgery. In recent years, efforts have been made on developing computerized dental X-ray image analysis systems for clinical usages. A novel framework for objective evaluation of automatic dental radiography analysis algorithms has been established under the auspices of the IEEE International Symposium on Biomedical Imaging 2015 Bitewing Radiography Caries Detection Challenge and Cephalometric X-ray Image Analysis Challenge. In this article, we present the datasets, methods and results of the challenge and lay down the principles for future uses of this benchmark. The main contributions of the challenge include the creation of the dental anatomy data repository of bitewing radiographs, the creation of the anatomical abnormality classification data repository of cephalometric radiographs, and the definition of objective quantitative evaluation for comparison and ranking of the algorithms. With this benchmark, seven automatic methods for analysing cephalometric X-ray image and two automatic methods for detecting bitewing radiography caries have been compared, and detailed quantitative evaluation results are presented in this paper. Based on the quantitative evaluation results, we believe automatic dental radiography analysis is still a challenging and unsolved problem. The datasets and the evaluation software will be made available to the research community, further encouraging future developments in this field. (http://www-o.ntust.edu.tw/~cweiwang/ISBI2015/)

In this paper, the authors describe an extension to an approach previously discussed for personalization of a natural language system in the automotive domain that allows reasoning under uncertainty with incomplete preference structures. Therefore, the concept of an "information stream" is defined as an underlying model for real-time recommendation learned from previous speech queries. The stream captures contextual data based on implicit feedback from the user's speech utterances. Furthermore, a formative user study is discussed. Each study iteration has been based on a prototype that allows the user to utter natural language queries in the restaurant domain. The system responds with a ranked list of restaurant recommendations in relation to the user's context. Several driving scenarios with varying contexts have been analyzed (e.g. weekday/weekend, route destinations, traffic). Users could inspect the result lists and indicate the most preferred item. In addition to quantitative data gained from this interaction, feedback on relevance of context features and on the UI concept was collected in a post-study interview for each iteration. Based on the study findings, we outline the contextual features found to be most relevant for speech-based interaction in automotive applications. These findings will be integrated into an existing hybrid recommendation model.

Recent work has shown that optical flow estimation can be formulated as a supervised learning task and can be successfully solved with convolutional networks. Training of the so-called FlowNet was enabled by a large synthetically generated dataset. The present paper extends the concept of optical flow estimation via convolutional networks to disparity and scene flow estimation. To this end, we propose three synthetic stereo video datasets with sufficient realism, variation, and size to successfully train large networks. Our datasets are the first large-scale datasets to enable training and evaluation of scene flow methods. Besides the datasets, we present a convolutional network for real-time disparity estimation that provides state-of-the-art results. By combining a flow and disparity estimation network and training it jointly, we demonstrate the first scene flow estimation with a convolutional network.

The development of space systems involves complex interdisciplinary systems engineering. The concurrent engineering (CE) approach has been successfully applied to the early design phase of space missions. To bridge the gap between the development phases and between the different domain experts, a model based system engineering (MBSE) approach is showing promising results. To support CE and MBSE during space mission development, the German Aerospace Center (DLR) has started developing a new tool called Virtual Satellite. It offers extended software support required by CE for inter-domain communication, data exchange, dependency analysis, on the fly data analysis, data consistency while maintaining a common system model based on the MBSE approach. However, the general issues of inter-domain communication and understanding still exist and may lead to misinterpretation. To overcome this problem it is intended to take advantage of interactive 3D visualization and Virtual Reality techniques to visualize the complex system model and, thus, provide a common understanding of the system model and the intrinsic domain knowledge. Furthermore, this will promote the experts to communicate their ideas and improve the visibility of potential design issues. The paper describes the efforts taken at DLR in this direction, architecture details and advantages of adopting these techniques into space mission development from the early design phase.