Informatik
Refine
Year of publication
Document Type
- Conference proceeding (83) (remove)
Language
- English (83)
Is part of the Bibliography
- yes (83)
Institute
- Informatik (83)
Publisher
- IEEE (83) (remove)
Among the multitude of software development processes available, hardly any is used by the book. Regardless of company size or industry sector, a majority of project teams and companies use customized processes that combine different development methods— so-called hybrid development methods. Even though such hybrid development methods are highly individualized, a common understanding of how to systematically construct synergetic practices is missing. In this paper, we make a first step towards devising such guidelines. Grounded in 1,467 data points from a large-scale online survey among practitioners, we study the current state of practice in process use to answer the question: What are hybrid development methods made of? Our findings reveal that only eight methods and few practices build the core of modern software development. This small set allows for statistically constructing hybrid development methods. Using an 85% agreement level in the participants’ selections, we provide two examples illustrating how hybrid development methods are characterized by the practices they are made of. Our evidence-based analysis approach lays the foundation for devising hybrid development methods.
In this paper we describe an interactive web-based visual analysis tool for Formula one races. It first provides an overview about all races on a yearly basis in a calendar-like representation. From this starting point, races can be selected and visually inspected in detail. We support a dynamic race position diagram as well as a more detailed lap times line plot for showing the drivers’ lap times in comparison. Many interaction techniques are supported like selections, filtering, highlighting, color coding, or details-on demand. We illustrate the usefulness of our visualization tool by applying it to a Formula one dataset while we describe the different dynamic visual racing patterns for a number of selected races and drivers.
Recognizing actions of humans, reliably inferring their meaning and being able to potentially exchange mutual social information are core challenges for autonomous systems when they directly share the same space with humans. Today’s technical perception solutions have been developed and tested mostly on standard vision benchmark datasets where manual labeling of sensory ground truth is a tedious but necessary task. Furthermore, rarely occurring human activities are underrepresented in such data leading to algorithms not recognizing such activities. For this purpose, we introduce a modular simulation framework which offers to train and validate algorithms on various environmental conditions. For this paper we created a dataset, containing rare human activities in urban areas, on which a current state of the art algorithm for pose estimation fails and demonstrate how to train such rare poses with simulated data only.
Engineers of the research project “Digital Product Life-Cycle” are using a graph-based design language to model all aspects of the product they are working on. This abstract model is the base for all further investigations, developments and implementations. In particular at early stages of development, collaborative decision making is very important. We propose a semantic augmented knowledge space by means of mixed reality technology, to support engineering teams. Therefore we present an interaction prototype consisting of a pico projector and a camera. In our usage scenario engineers are augmenting different artefacts in a virtual working environment. The concept of our prototype contains both an interaction and a technical concept. To realise implicit and natural interactions, we conducted two prototype tests: (1) A test with a low-fidelity prototype and (2) a test by using the method Wizard of Oz. As a result, we present a prototype with interaction selection using augmentation spotlighting and an interaction zoom as a semantic zoom.
This research addresses the question of why employees use enterprise social networks (ESN). Against the background of technology acceptance research, we propose an extended unified theory of acceptance and use of technology (UTAUT) model, adapt it to an ESN context, and test our model against data from ESN users of large and medium-sized enterprises. We use partial least squares structural equation modeling to gain insights into the determinants of ESN use. This paper contributes to ESN acceptance research by evaluating a model containing determinants of ESN use. It also examines the effects of determinants on five different usage dimensions of ESN. The results reveal that facilitating conditions are the main driver of ESN use while the impact of intention to use is comparably small. Implications for theory and practice are discussed.
Transaction processing is of growing importance for mobile computing. Booking tickets, flight reservation, banking, ePayment, and booking holiday arrangements are just a few examples for mobile transactions. Due to temporarily disconnected situations the synchronisation and consistent transaction processing are key issues. Serializability is a too strong criteria for correctness when the semantics of a transaction is known. We introduce a transaction model that allows higher concurrency for a certain class of transactions defined by its semantic. The transaction results are ”escrow serializable” and the synchronisation mechanism is non-blocking. Experimental implementation showed higher concurrency, transaction throughput, and less resources used than common locking or optimistic protocols.
IT environments that consist of a very large number of rather small structures like microservices, Internet of Things (IoT) components, or mobility systems are emerging to support flexible and agile products and services in the age of digital transformation. Biological metaphors of living and adaptable ecosystems with service-oriented enterprise architectures provide the foundation for self-optimizing, resilient run-time environments and distributed information systems. We are extending Enterprise Architecture (EA) methodologies and models that cover a high degree of heterogeneity and distribution to support the digital transformation and related information systems with micro-granular architectures. Our aim is to support flexibility and agile transformation for both IT and business capabilities within adaptable digital enterprise architectures. The present research paper investigates mechanisms for integrating Microservice Architectures (MSA) by extending original enterprise architecture reference models with elements for more flexible architectural metamodels and EA-mini-descriptions.
AI technologies such as deep learning provide promising advances in many areas. Using these technologies, enterprises and organizations implement new business models and capabilities. In the beginning, AI-technologies have been deployed in an experimental environment. AI-based applications have been created in an ad-hoc manner and without methodological guidance or engineering approach. Due to the increasing importance of AI-technologies, however, a more structured approach is necessary that enable the methodological engineering of AI-based applications. Therefore, we develop in this paper first steps towards methodological engineering of AI-based applications. First, we identify some important differences between the technological foundations of AI- technologies, in particular deep learning, and traditional information technologies. Then we create a framework that enables to engineer AI-applications using four steps: identification of an AI-application type, sub-type identification, lifecycle phase, and definition of details. The introduced framework considers that AI-applications use an inductive approach to infer knowledge from huge collections and streams of data. It not only enables the rapid development of AI-application but also the efficient sharing of knowledge on AI-applications.
Continuous refactoring is necessary to maintain source code quality and to cope with technical debt. Since manual refactoring is inefficient and error prone, various solutions for automated refactoring have been proposed in the past. However, empirical studies have shown that these solutions are not widely accepted by software developers and most refactorings are still performed manually. For example, developers reported that refactoring tools should support functionality for reviewing changes. They also criticized that introducing such tools would require substantial effort for configuration and integration into the current development environment.
In this paper, we present our work towards the Refactoring-Bot, an autonomous bot that integrates into the team like a human developer via the existing version control platform. The bot automatically performs refactorings to resolve code smells and presents the changes to a developer for asynchronous review via pull requests. This way, developers are not interrupted in their workflow and can review the changes at any time with familiar tools. Proposed refactorings can then be integrated into the code base via the push of a button. We elaborate on our vision, discuss design decisions, describe the current state of development, and give an outlook on planned development and research activities.
Current approaches for enterprise architecture lack analytical instruments for cyclic evaluations of business and system architectures in real business enterprise system environments. This impedes the broad use of enterprise architecture methodologies. Furthermore, the permanent evolution of systems desynchronizes quickly model representation and reality. Therefore we are introducing an approach for complementing the existing top-down approach for the creation of enterprise architecture with a bottom approach. Enterprise Architecture Analytics uses the architectural information contained in many infrastructures to provide architectural information. By applying Big Data technologies it is possible to exploit this information and to create architectural information. That means, Enterprise Architectures may be discovered, analyzed and optimized using analytics. The increased availability of architectural data also improves the possibilities to verify the compliance of Enterprise Architectures. Architectural decisions are linked to clustered architecture artifacts and categories according to a holistic EAM Reference Architecture with specific architecture metamodels. A special suited EAM Maturity Framework provides the base for systematic and analytics supported assessments of architecture capabilities.
While the concepts of object-oriented antipatterns and code smells are prevalent in scientific literature and have been popularized by tools like SonarQube, the research field for service-based antipatterns and bad smells is not as cohesive and organized. The description of these antipatterns is distributed across several publications with no holistic schema or taxonomy. Furthermore, there is currently little synergy between documented antipatterns for the architectural styles SOA and Microservices, even though several antipatterns may hold value for both. We therefore conducted a Systematic Literature Review (SLR) that identified 14 primary studies. 36 service-based antipatterns were extracted from these studies and documented with a holistic data model. We also categorized the antipatterns with a taxonomy and implemented relationships between them. Lastly, we developed a web application for convenient browsing and implemented a GitHub-based repository and workflow for the collaborative evolution of the collection. Researchers and practitioners can use the repository as a reference, for training and education, or for quality assurance.
Early reduction of risks in a startup or an innovation project is highly important. Appropriate means for risk reduction, such as testing business models with different kinds of experiments exist. However, deciding what to test and how to select the right test, is challenging for many startups and innovation projects. This article presents the so-called Business Experiments Navigator (BEN), a toolkit to assist startup and innovation processes. It compliments other tools such as the Business Model Canvas or the Lean Startup process. The main contribution of BEN is to bridge the gap between the riskiest assumptions of a business model and the multitude of available testing techniques by providing assumption templates. The Business Experiments Navigator has been validated in several workshops. Results show that it creates awareness among the workshop participants that a business model is based on assumptions which impose risks and need to be validated. Further, users of BEN were able to identify relevant assumptions and map different kinds of assumptions to appropriate testing techniques. The process applied in the workshops, as well as the assumption templates, helped the participants understand the main concepts and transfer their learnings, to their own business ideas.
Software process improvement (SPI) is around for decades, but it is a critically discussed topic. In several waves, different aspects of SPI have been discussed in the past, e.g., large scale company-level SPI programs, maturity models, success factors, and in-project SPI. It is hard to find new streams or a consensus in the community, but there is a trend coming along with agile and lean software development. Apparently, practitioners reject extensive and prescriptive maturity models and move towards smaller, faster and continuous project-integrated SPI. Based on data from two survey studies conducted in Germany (2012) and Europe (2016), we analyze the process customization for projects and practices for implementing SPI in the participating companies. Our findings indicate that, even in regulated industry sectors, companies increasingly adopt in-project SPI activities, primarily with the goal to continuously optimize specific processes. Therefore, with this paper, we want to stimulate a discussion on how to evolve traditional SPI towards a continuous learning environment.
To evaluate the quality of a person´s sleep it is essential to identify the sleep stages and their durations. Currently, the gold standard in terms of sleep analysis is overnight polysomnography (PSG), during which several techniques like EEG (eletroencephalogram), EOG (electrooculogram), EMG (electromyogram), ECG (electrocardiogram), SpO2 (blood oxygen saturation) and for example respiratory airflow and respiratory effort are recorded. These expensive and complex procedures, applied in sleep laboratories, are invasive and unfamiliar for the subjects and it is a reason why it might have an impact on the recorded data. These are the main reasons why low-cost home diagnostic systems are likely to be advantageous. Their aim is to reach a larger population by reducing the number of parameters recorded. Nowadays, many wearable devices promise to measure sleep quality using only the ECG and body-movement signals. This work presents an android application developed in order to proof the accuracy of an algorithm published in the sleep literature. The algorithm uses ECG and body movement recordings to estimate sleep stages. The pre-recorded signals fed into the algorithm have been taken from physionet1 online database. The obtained results have been compared with those of the standard method used in PSG. The mean agreement ratios between the sleep stages REM, Wake, NREM-1, NREM-2 and NREM-3 were 38.1%, 14%, 16%, 75% and 54.3%.
The respiratory rate is a vital sign indicating breathing illness. It is necessary to analyze the mechanical oscillations of the patient's body arising from chest movements. An inappropriate holder on which the sensor is mounted, or an inappropriate sensor position is some of the external factors which should be minimized during signal registration. This paper considers using a non-invasive device placed under the bed mattress and evaluates the respiratory rate. The aim of the work is the development of an accelerometer sensor holder for this system. The normal and deep breathing signals were analyzed, corresponding to the relaxed state and when taking deep breaths. The evaluation criterion for the holder's model is its influence on the patient's respiratory signal amplitude for each state. As a result, we offer a non-invasive system of respiratory rate detection, including the mechanical component providing the most accurate values of mentioned respiratory rate.
Recognizing human actions is a core challenge for autonomous systems as they directly share the same space with humans. Systems must be able to recognize and assess human actions in real-time. To train the corresponding data-driven algorithms, a significant amount of annotated training data is required. We demonstrate a pipeline to detect humans, estimate their pose, track them over time and recognize their actions in real-time with standard monocular camera sensors. For action recognition, we transform noisy human pose estimates in an image like format we call Encoded Human Pose Image (EHPI). This encoded information can further be classified using standard methods from the computer vision community. With this simple procedure, we achieve competitive state-of-the-art performance in pose based action detection and can ensure real-time performance. In addition, we show a use case in the context of autonomous driving to demonstrate how such a system can be trained to recognize human actions using simulation data.
Serverless computing is an emerging cloud computing paradigm with the goal of freeing developers from resource management issues. As of today, serverless computing platforms are mainly used to process computations triggered by events or user requests that can be executed independently of each other. These workloads benefit from on-demand and elastic compute resources as well as per-function billing. However, it is still an open research question to which extent parallel applications, which comprise most often complex coordination and communication patterns, can benefit from serverless computing.
In this paper, we introduce serverless skeletons for parallel cloud programming to free developers from both parallelism and resource management issues. In particular, we investigate on the well known and widely used farm skeleton, which supports the implementation of a wide range of applications. To evaluate our concepts, we present a prototypical development and runtime framework and implement two applications based on our framework: Numerical integration and hyperparameter optimization - a commonly applied technique in machine learning. We report on performance measurements for both applications and discuss
the usefulness of our approach.
OpenAPI, WADL, RAML, and API Blueprint are popular formats for documenting Web APIs. Although these formats are in general both human and machine-readable, only the part of the format describing the syntax of a Web API is machine-understandable. Descriptions, which explain the meaning and purpose of Web API elements, are embedded as natural language text snippets into documents and target human readers but not machines. To enable machines to read and process these state-of-practice Web API documentation, we propose a Transformer model that solves the generic task of identifying a Web API element within a syntax structure that matches a natural language query. For our first prototype, we focus on the Web API integration task of matching output with input parameters and fined-tuned a pre-trained CodeBERT model to the downstream task of question answering with samples from 2,321 OpenAPI documentation. We formulate the original question answering problem as a multiple choice task: given a semantic natural language description of an output parameter (question) and the syntax of the input schema (paragraph), the model chooses the input parameter (answer) in the schema that best matches the description. The paper describes the data preparation, tokenization, and fine-tuning process as well as discusses possible applications of our model as part of a recommender system. Furthermore, we evaluate the generalizability and the robustness of our fine-tuned model, with the result that it achieves an accuracy of 81.46% correctly chosen parameters.
In the present paper we demonstrate the novel technique to apply the recently proposed approach of In-Place Appends – overwrites on Flash without a prior erase operation. IPA can be applied selectively: only to DB-objects that have frequent and relatively small updates. To do so we couple IPA to the concept of NoFTL regions, allowing the DBA to place update-intensive DB-objects into special IPA-enabled regions. The decision about region configuration can be (semi-)automated by an advisor analyzing DB-log files in the background.
We showcase a Shore-MT based prototype of the above approach, operating on real Flash hardware. During the demonstration we allow the users to interact with the system and gain hands-on experience under different demonstration scenarios.
Methods based exclusively on heart rate hardly allow to differentiate between physical activity, stress, relaxation, and rest, that is why an additional sensor like activity/movement sensor added for detection and classification. The response of the heart to physical activity, stress, relaxation, and no activity can be very similar. In this study, we can observe the influence of induced stress and analyze which metrics could be considered for its detection. The changes in the Root Mean Square of the Successive Differences provide us with information about physiological changes. A set of measurements collecting the RR intervals was taken. The intervals are used as a parameter to distinguish four different stages. Parameters like skin conductivity or skin temperature were not used because the main aim is to maintain a minimum number of sensors and devices and thereby to increase the wearability in the future.
As production workspaces become more mobile and dynamic it becomes increasingly important to reliably monitor the overall state of the environment. Therein manipulators or other robotic systems likely have to be able to act autonomously together with humans and other systems within a joint workspace. Such interactions require that all components in non-stationary environments are able to perceive the state relative to each other. As vision-sensors provide a rich source of information to accomplish this, we present RoPose, a convolutional neural network (CNN) based approach, to estimate the two dimensional joint configuration of a simulated industrial manipulator from a camera image. This pose information can further be used by a novel targetless calibration setup to estimate the pose of the camera relative to the manipulator’s space. We present a pipeline to automatically generate synthetic training data and conclude with a discussion of the potential usage of the same pipeline to acquire real image datasets of physically existent robots.
RoPose-Real: real world dataset acquisition for data-driven industrial robot arm pose estimation
(2019)
It is necessary to employ smart sensory systems in dynamic and mobile workspaces where industrial robots are mounted on mobile platforms. Such systems should be aware of flexible and non-stationary workspaces and able to react autonomously to changing situations. Building upon our previously presented RoPose-system, which employs a convolutional neural network architecture that has been trained on pure synthetic data to estimate the kinematic chain of an industrial robot arm system, we now present RoPose-Real. RoPose-Real extends the prior system with a comfortable and targetless extrinsic calibration tool, to allow for the production of automatically annotated datasets for real robot systems. Furthermore, we use the novel datasets to train the estimation network with real world data. The extracted pose information is used to automatically estimate the observing sensor pose relative to the robot system. Finally we evaluate the performance of the presented subsystems in a real world robotic scenario.
Significant advances have been achieved in mobile robot localization and mapping in dynamic environments, however these are mostly incapable of dealing with the physical properties of automotive radar sensors. In this paper we present an accurate and robust solution to this problem, by introducing a memory efficient cluster map representation. Our approach is validated by experiments that took place on a public parking space with pedestrians, moving cars, as well as different parking configurations to provide a challenging dynamic environment. The results prove its ability to reproducibly localize our vehicle within an error margin of below 1% with respect to ground truth using only point based radar targets. A decay process enables our map representation to support local updates.
In this paper, an approach is introduced how reinforcement learning can be used to achieve interoperability between heterogeneous Internet of Things (IoT) components. More specifically, we model an HTTP REST service as a Markov Decision Process and adapt Q-Learning to the properties of REST so that an agent in the role of an HTTP REST client can learn the semantics of the service and, especially an optimal sequence of service calls to achieve an application specific goal. With our approach, we want to open up and facilitate a discussion in the community, as we see the key for achieving interoperability in IoT by the utilization of artificial intelligence techniques.
This document presents an algorithm for a nonobtrusive recognition of Sleep/Wake states using signals derived from ECG, respiration, and body movement captured while lying in a bed. As a core mathematical base of system data analytics, multinomial logistic regression techniques were chosen. Derived parameters of the three signals are used as the input for the proposed method. The overall achieved accuracy rate is 84% for Wake/Sleep stages, with Cohen’s kappa value 0.46. The presented algorithm should support experts in analyzing sleep quality in more detail. The results confirm the potential of this method and disclose several ways for its improvement.
Context: The software-intensive business is characterized by increasing market dynamics, rapid technological changes, and fast-changing customer behaviors. Organizations face the challenge of moving away from traditional roadmap formats to an outcome-oriented approach that focuses on delivering value to the customer and the business. An important starting point and a prerequisite for creating such outcome-oriented roadmaps is the development of a product vision to which internal and external stakeholders can be aligned. However, the process of creating a product vision is little researched and understood.
Objective: The goal of this paper is to identify lessons-learned from product vision workshops, which were conducted to develop outcome-oriented product roadmaps.
Method: We conducted a multiple-case study consisting of two different product vision workshops in two different corporate contexts.
Results: Our results show that conducting product vision workshops helps to create a common understanding among all stakeholders about the future direction of the products. In addition, we identified key organizational aspects that contribute to the success of product vision workshops, including the participation of employees from functionally different departments.
Product roadmaps in the new mobility domain: state of the practice and industrial experiences
(2021)
Context: The New Mobility industry is a young market that includes high market dynamics and is therefore associated with a high degree of uncertainty. Traditional product roadmapping approaches such a detailed planning of features over a long-time horizon typically fail in such environments. For this reason, companies that are active in the field of New Mobility are faced with the challenge of keeping their product roadmaps reliable for stakeholders while at the same time being able to react flexibly to changing market requirements.
Objective: The goal of this paper is to identify the state of practice regarding product roadmapping of New Mobility companies. In addition, the related challenges within the product roadmapping process as well as the success factors to overcome these challenges will be highlighted.
Method: We conducted semi-structured expert interviews with 8 experts (7 German company and one Finnish company) from the field of New Mobility and performed a content analysis.
Results: Overall the results of the study showed that the participating companies are aware of the requirements that the New Mobility sector entails. Therefore, they exhibit a high level of maturity in terms of product roadmapping. Nevertheless, some aspects were revealed that pose specific challenges for the participating companies. One major challenge, for example, is that New Mobility in terms of public clients is often a tender business with non-negotiable product requirements. Thus, the product roadmap can be significantly influenced from the outside. As factors for a successful product roadmapping mainly soft factors such as trust between all people involved in the product development process and transparency throughout the entire roadmapping process were mentioned.
Context: Companies in highly dynamic markets increasingly struggle with their ability to plan product development and to create reliable roadmaps. A main reason is the decreasing lack of predictability of markets, technologies, and customer behaviors. New approaches for product roadmapping seem to be necessary in order to cope with today's highly dynamic conditions. Little research is available with respect to such new approaches. Objective: In order to better understand the state of the art and to identify research gaps, this article presents a review of the scientific literature with respect to product roadmapping. Method: We performed a systematic literature review (SLR) with respect to identify papers in the field of computer science. Results: After filtering, the search resulted in a set of 23 relevant papers. The identified papers focus on different aspects such as roadmap types, processes for creating and updating roadmaps, problems and challenges with roadmapping, approaches to visualize roadmaps, generic frameworks and specific aspects such as the combination of roadmaps with business modeling. Overall, the scientific literature covers many important aspects of roadmapping but does provide only little knowledge on how to create product roadmaps under highly dynamic conditions. Research gaps address, for instance, the inclusion of goals or outcomes into product roadmaps, the alignment of a roadmap with a product vision, and the inclusion of product discovery activities in product roadmaps. In addition, the transformation from traditional roadmapping processes to new ways of roadmapping is not sufficiently addressed in the scientific literature.
Product roadmaps are an important tool in product development. They provide direction, enable consistent development in relation to a product vision and support communication with relevant stakeholders. There are many different formats for product roadmaps, but they are often based on the assumption that the future is highly predictable. However, especially software-intensive businesses are faced with increasing market dynamics, rapidly evolving technologies and changing user expectations. As a result, many organizations are wondering what roadmap format is appropriate for them and what components it should have to deal with an unpredictable future. Objectives: To gain a better understanding of the formats of product roadmaps and their components, this paper aims to identify suitable formats for the development and handling of product roadmaps in dynamic and uncertain markets. Method: We performed a grey literature review (GLR) according to the guidelines from Garousi. Results: A Google search identified 426 articles, 25 of which were included in this study. First, various components of the roadmap were identified, especially the product vision, themes, goals, outcomes and outputs. In addition, various product roadmap formats were discovered, such as feature-based, goal-oriented, outcome-driven and a theme-based roadmap. The roadmap components were then assigned to the various product roadmap formats. This overview aims at providing initial decision support for companies to select a suitable product roadmap format and adapt it to their own needs.
In recent years companies have faced challenges by high market dynamics, rapidly evolving technologies and shifting user expectations. Together with the adaption of lean and agile practices, it is increasingly difficult to predict upfront which products, features or services will satisfy the needs of the customers and the organization. Currently, many new products fail to produce a significant financial return. One reason is that companies are not doing enough product discovery activities. Product discovery aims at tackling the various risks before the implementation of a product starts. The academic literature only provides little guidance for conducting product discovery in practice. Objective: In order to gain a better understanding of product discovery activities in practice, this paper aims at identifying motivations, approaches, challenges, risks, and pitfalls of product discovery reported in the grey literature. Method: We performed a grey literature review (GLR) according to the guidelines to Garousi et al. Results: The study shows that the main motivation for conducting product discovery activities is to reduce the uncertainty to a level that makes it possible to start building a solution that provides value for the customers and the business. Several product discovery approaches are reported in the grey literature which include different phases such as alignment, problem exploration, ideation, and validation. Main challenges are, among others, the lack of clarity of the problem to be solved, the prescription of concrete solutions through management or experts, and the lack of cross-functional collaboration.
Additive manufacturing (AM) is a promising manufacturing method for many industrial sectors. For this application, industrial requirements such as high production volumes and coordinated implementation must be taken into account. These tasks of the internal handling of production facilities are carried out by the Production Planning and Control (PPC) information system. A key factor in the planning and scheduling is the exact calculation of manufacturing times. For this purpose we investigate the use of Machine Learning (ML) for the prediction of manufacturing times of AM facilities.
Human pose estimation (HPE) is integral to scene understanding in numerous safety-critical domains involving human-machine interaction, such as autonomous driving or semi-automated work environments. Avoiding costly mistakes is synonymous with anticipating failure in model predictions, which necessitates meta-judgments on the accuracy of the applied models. Here, we propose a straightforward human pose regression framework to examine the behavior of two established methods for simultaneous aleatoric and epistemic uncertainty estimation: maximum a-posteriori (MAP) estimation with Monte-Carlo variational inference and deep evidential regression (DER). First, we evaluate both approaches on the quality of their predicted variances and whether these truly capture the expected model error. The initial assessment indicates that both methods exhibit the overconfidence issue common in deep probabilistic models. This observation motivates our implementation of an additional recalibration step to extract reliable confidence intervals. We then take a closer look at deep evidential regression, which, to our knowledge, is applied comprehensively for the first time to the HPE problem. Experimental results indicate that DER behaves as expected in challenging and adverse conditions commonly occurring in HPE and that the predicted uncertainties match their purported aleatoric and epistemic sources. Notably, DER achieves smooth uncertainty estimates without the need for a costly sampling step, making it an attractive candidate for uncertainty estimation on resource-limited platforms.
We present a multitask network that supports various deep neural network based pedestrian detection functions. Besides 2D and 3D human pose, it also supports body and head orientation estimation based on full body bounding box input. This eliminates the need for explicit face recognition. We show that the performance of 3D human pose estimation and orientation estimation is comparable to the state-of-the-art. Since very few data sets exist for 3D human pose and in particular body and head orientation estimation based on full body data, we further show the benefit of particular simulation data to train the network. The network architecture is relatively simple, yet powerful, and easily adaptable for further research and applications.
Massive data transfers in modern data intensive systems resulting from low data-locality and data-to-code system design hurt their performance and scalability. Near-data processing (NDP) and a shift to code-to-data designs may represent a viable solution as packaging combinations of storage and compute elements on the same device has become viable.
The shift towards NDP system architectures calls for revision of established principles. Abstractions such as data formats and layouts typically spread multiple layers in traditional DBMS, the way they are processed is encapsulated within these layers of abstraction. The NDP-style processing requires an explicit definition of cross-layer data formats and accessors to ensure in-situ executions optimally utilizing the properties of the underlying NDP storage and compute elements. In this paper, we make the case for such data format definitions and investigate the performance benefits under NoFTL-KV and the COSMOS hardware platform.
Multi-versioning and MVCC are the foundations of many modern DBMSs. Under mixed workloads and large datasets, the creation of the transactional snapshot can become very expensive, as long-running analytical transactions may request old versions, residing on cold storage, for reasons of transactional consistency. Furthermore, analytical queries operate on cold data, stored on slow persistent storage. Due to the poor data locality, snapshot creation may cause massive data transfers and thus lower performance. Given the current trend towards computational storage and near-data processing, it has become viable to perform such operations in-storage to reduce data transfers and improve scalability. neoDBMS is a DBMS designed for near-data processing and computational storage. In this paper, we demonstrate how neoDBMS performs snapshot computation in-situ. We showcase different interactive scenarios, where neoDBMS outperforms PostgreSQL 12 by up to 5×.
In the present tutorial we perform a cross-cut analysis of database storage management from the perspective of modern storage technologies. We argue that neither the design of modern DBMS, nor the architecture of modern storage technologies are aligned with each other. Moreover, the majority of the systems rely on a complex multi-layer and compatibility oriented storage stack. The result is needlessly suboptimal DBMS performance, inefficient utilization, or significant write amplification due to outdated abstractions and interfaces. In the present tutorial we focus on the concept of native storage, which is storage operated without intermediate abstraction layers over an open native storage interface and is directly controlled by the DBMS.
The internet of things, enterprise social networks, adaptive case management, mobility systems, analytics for big data, and cloud environments are emerging to support smart connected i.e. digital products and services and the digital transformation. Biological metaphors for living and adaptable ecosystems are currently providing the logical foundation for resilient run-time environments with serviceoriented digitization architectures and for self-optimizing intelligent business services and related distributed information systems. We are investigating mechanisms for flexible adaptation and evolution of information systems with digital architecture in the context of the ongoing digital transformation. The goal is to support flexible and agile transformations for both business and related information systems through adaptation and dynamical evolution of their digital architectures. The present research paper investigates mechanisms of decision analytics for digitization architectures, putting a spotlight to internet of things micro-granular architectures, by extending original enterprise architecture reference models with digitization architectures and their multi-perspective architectural decision management.
To remain competitive in a fast changing environment, many companies started to migrate their legacy applications towards a Microservices architecture. Such extensive migration processes require careful planning and consideration of implications and challenges likewise. In this regard, hands-on experiences from industry practice are still rare. To fill this gap in scientific literature, we contribute a qualitative study on intentions, strategies, and challenges in the context of migrations to Microservices. We investigated the migration process of 14 systems across different domains and sizes by conducting 16 in-depth interviews with software professionals from 10 companies. Along with a summary of the most important findings, we present a separate discussion of each case. As primary migration drivers, maintainability and scalability were identified. Due to the high complexity of their legacy systems, most companies preferred a rewrite using current technologies over splitting up existing code bases. This was often caused by the absence of a suitable decomposition approach. As such, finding the right service cut was a major technical challenge, next to building the necessary expertise with new technologies. Organizational challenges were especially related to large, traditional companies that simultaneously established agile processes. Initiating a mindset change and ensuring smooth collaboration between teams were crucial for them. Future research on the evolution of software systems can in particular profit from the individual cases presented.
Microservices are a topic driven mainly by practitioners and academia is only starting to investigate them. Hence, there is no clear picture of the usage of Microservices in practice. In this paper, we contribute a qualitative study with insights into industry adoption and implementation of Microservices. Contrary to existing quantitative studies, we conducted interviews to gain a more in-depth understanding of the current state of practice. During 17 interviews with software professionals from 10 companies, we analyzed 14 service-based systems. The interviews focused on applied technologies, Microservices characteristics, and the perceived influence on software quality. We found that companies generally rely on well established technologies for service implementation, communication, and deployment. Most systems, however, did not exhibit a high degree of technological diversity as commonly expected with Microservices. Decentralization and product character were different for systems built for external customers. Applied DevOps practices and automation were still on a mediocre level and only very few companies strictly followed the you build it, you run it principle. The impact of Microservices on software quality was mainly rated as positive. While maintainability received the most positive mentions, some major issues were associated with security. We present a description of each case and summarize the most important findings of companies across different domains and sizes. Researchers may build upon our findings and take them into account when designing industry-focused methods.
This research evaluates current measurement scales for ambidexterity and proposes a new approach for the measurement of this important construct. We argue that current measurement approaches may be unsuitable to capture the concept of ambidexterity. Through a systematic scale development process, we derive a measurement scale with dual items that simultaneously refer to both dimensions, exploitation and exploration, thus reflecting the true nature of ambidexterity. An extensive pre-test with 39 executives suggests that our scale is suitable for capturing ambidexterity. Our measurement model enhances conceptual clarity of ambidexterity and can serve as a base for future investigations of the concept.
Learning to translate between real world and simulated 3D sensors while transferring task models
(2019)
Learning-based vision tasks are usually specialized on the sensor technology for which data has been labeled. The knowledge of a learned model is simply useless when it comes to data which differs from the data on which the model has been initially trained or if the model should be applied to a totally different imaging or sensor source. New labeled data has to be acquired on which a new model can be trained. Depending on the sensor, this can even get more complicated when the sensor data becomes more abstract and hard to be interpreted and labeled by humans. To enable reuse of models trained for a specific task across different sensors minimizes the data acquisition effort. Therefore, this work focuses on learning sensor models and translating between them, thus aiming for sensor interoperability. We show that even for the complex task of human pose estimation from 3D depth data recorded with different sensors, i.e. a simulated and a Kinect 2TM depth sensor, human pose estimation can greatly improve by translating between sensor models without modifying the original task model. This process especially benefits sensors and applications for which labels and models are difficult if at all possible to retrieve from raw sensor data.
On the way to achieving higher degrees of autonomy for vehicles in complicated, ever changing scenarios, the localization problem poses a very important role. Especially the Simultaneous Localization and Mapping (SLAM) problem has been studied greatly in the past. For an autonomous system in the real world, we present a very cost-efficient, robust and very precise localization approach based on GraphSLAM and graph optimization using radar sensors. We are able to prove on a dynamically changing parking lot layout that both mapping and localization accuracy are very high. To evaluate the performance of the mapping algorithm, a highly accurate ground truth map generated from a total station was used. Localization results are compared to a high precision DGPS/INS system. Utilizing these methods, we can show the strong performance of our algorithm.
In any autonomous driving system, the map for localization plays a vital part that is often underestimated. The map describes the world around the vehicle outside of the sensor view and is a main input into the decision making process in highly complicated scenarios. Thus there are strict requirements towards the accuracy and timeliness of the map. We present a robust and reliable approach towards crowd based mapping using a GraphSLAM framework based on radar sensors. We show on a parking lot that even in dynamically changing environments, the localization results are very accurate and reliable even in unexplored terrain without any map data. This can be achieved by collaborative map updates from multiple vehicles. To show these claims experimentally, the Joint Graph Optimization is compared to the ground truth on an industrial parking space. Mapping performance is evaluated using a dense map from a total station as reference and localization results are compared with a deeply coupled DGPS/INS system.
Analysis and planning of Enterprise Architectures (EA) is a complex task for stakeholders. The change of one architecture element has impact on multiple other elements because of manifold relationships and interactions between them. The interactive cockpit approach presented in this paper supports stakeholders planning and analyzing EAs and to tackle the intrinsic complexity. This approach supplies a cockpit with multiple viewpoints to put relevant information side-by-side without losing the context combined with interaction functionality. In this paper, we develop such cockpit starting with relevant use cases, describing a potential design based on well-established foundations in EA modeling, and outline an exemplary usage scenario.
Context: The manufacturing industry is facing a transformation with regard to Industry 4.0 (I4). A transformation towards full automation of production including a multitude of innovations is necessary. Startups and entrepreneurial processes can support such a transformation as has been shown in other industries. However, I4 has some specifics, so it is unclear how entrepreneurship can be adapted in I4. Understanding these specifics is important to develop suitable training programs for I4 startups and to accelerate the transformation.
Objective: This study identifies and outlines the essential characteristics and constraints of entrepreneurial processes in I4.
Method: 14 semi-structured interviews were conducted with experts in the field of I4 entrepreneurship. The interviews were analysed and categorized by qualitative analyses.
Results: The interviews revealed several characteristics of I4 that have a significant impact on the various phases of the entrepreneurial process. Examples of such specifics include the difficult access to customers, the necessary deep understanding of the customer and the domain, the difficulty of testing risky assumptions, and the complex development and productization of solutions. The complexity of hardware and software components, cost structures, and necessary customer-specific customizations affect the scalability of I4 startups. These essential characteristics also require specialised skills and resources from I4 startups.
Incubators in multinational corporations : development of a corporate incubator operator model
(2017)
This paper analyzes the components of a corporate incubator operator model in multinational companies. Thereby, three relevant phases were identified: pre incubation, incubation, and exit. Each phase contains different criteria that represent critical success factors for a corporate incubator, which are based on theoretical findings and lessons learned from practice. During the pre-incubation phase companies should define their need for a corporate incubator, the origin of ideas and the selection criteria for incubator tenants. The actual phase of incubation refers to the incubator program, which should be flexible with respect to each tenant. Furthermore, resource allocation plays an important role during the incubator program. Exit options after a successful incubation differ according to internal ideas and external start-ups, as well as the objective of the incubator. The research is based on a comprehensive screening of existing incubator literature and a qualitative content analysis of statements from eight experts of international corporate incubators.
Modern web-based applications are often built as multi-tier architecture using persistence middleware. Middleware technology providers recommend the use of Optimistic Concurrency Control (OCC) mechanism to avoid the risk of blocked resources. However, most vendors of relational database management systems implement only locking schemes for concurrency control. As consequence a kind of OCC has to be implemented at client or middleware side.
A simple Row Version Verification (RVV) mechanism has been proposed to implement an OCC at client side. For performance reasons the middleware uses buffers (cache) of its own to avoid network traffic and possible disk I/O. This caching however complicates the use of RVV because the data in the middleware cache may be stale (outdated). We investigate various data access technologies, including the new Java Persistence API (JPA) and Microsoft’s LINQ technologies for their ability to use the RVV programming discipline.
The use of persistence middleware that tries to relieve the programmer from the low level transaction programming turns out to even complicate the situation in some cases.Programmed examples show how to use SQL data access patterns to solve the problem.
Free-floating e-scooter sharing is an upcoming trend in mobility, which has been spreading since 2015 in various German cities. Unlike the more scientifically explorend car sharing, the usage patterns and behaviors of e-scooter sharing customers are yet to be analyzed. This presumably discovers better ways to attract customers as well as adaptions of the business model in order to increase scooter utilization and therefore the profit of the e-scooter providers. As most of the customer's journey, from registration to scooter reservation and the ride itself, is digitally traceable, large datasets are available allowing for understanding of customers' needs and motivations. Based on these datasets of an e-scooter provider operating in a big German city we propose a customer clustering that identifies four different customer segments, which enables multiple conclusions to be drawn for business development and improving the problem-solution fit of the e-scooter sharing model.
Measuring cardiorespiratory parameters in sleep, using non-contact sensors and the Ballistocardiography technique has received much attention due to the low-cost, unobtrusive, and non-invasive method. Designing a user-friendly, simple-to-use, and easy-to-deployment preserving less error-prone remains open and challenging due to the complex morphology of the signal. In this work, using four forcesensitive resistor sensors, we conducted a study by designing four distributions of sensors, in order to simplify the complexity of the system by identifying the region of interest for heartbeat and respiration measurement. The sensors are deployed under the mattress and attached to the bed frame without any interference with the subjects. The four distributions are combined in two linear horizontal, one linear vertical, and one square, covering the influencing region in cardiorespiratory activities. We recruited 4 subjects and acquired data in four regular sleeping positions, each for a duration of 80 seconds. The signal processing was performed using discrete wavelet transform bior 3.9 and smooth level of 4 as well as bandpass filtering. The results indicate that we have achieved the mean absolute error of 2.35 and 4.34 for respiration and heartbeat, respectively. The results recommend the efficiency of a triangleshaped structure of three sensors for measuring heartbeat and respiration parameters in all four regular sleeping positions.
How to prioritize your product roadmap when everything feels important: a grey literature review
(2021)
Context: A key factor in achieving product success is to identify what and in which order outputs must be launched in order to deliver the most value to the customer and the business. Therefore, a well-established process to discover and prioritize the content of the product roadmap in the right way is crucial for the success of a company. However, most companies prioritize their product roadmap items based on opinions of experts or the management. Additionally, increasing market dynamics, rapidly evolving technologies and fast changing customer behavior complicate the conduction of the prioritization process. Therefore, many companies are struggling to finding and establishing suitable techniques for prioritizing their product roadmap.
Objective: In order to gain a better understanding of the prioritization process in a dynamic and uncertain market environment, this paper aims to identify suitable techniques for the prioritization in such environments.
Method: We conducted a Grey Literature Review according to the guidelines of Garousi et al.
Results: 18 techniques for the prioritization of the product roadmap could be identified. 15 techniques are primarily used to prioritize outputs by considering factors such as the expected impact or effort. Two technique are most suitable for prioritizing risky assumptions that need to be validated and one technique focuses on the prioritization of outcomes. All techniques have in common that they should be conducted as cross-functional team activity in order to include different perspectives in the prioritization process.