Refine
Year of publication
- 2018 (37) (remove)
Document Type
- Conference proceeding (33)
- Journal article (4)
Language
- English (37)
Has full text
- yes (37)
Is part of the Bibliography
- yes (37)
Institute
- Technik (22)
- Informatik (11)
- ESB Business School (4)
Publisher
- IEEE (37) (remove)
SLAM systems are mainly applied for robot navigation while research on feasibility for motion planning with SLAM for tasks like bin-picking, is scarce. Accurate 3D reconstruction of objects and environments is important for planning motion and computing optimal gripper pose to grasp objects. In this work, we propose the methods to analyze the accuracy of a 3D environment reconstructed using a LSD-SLAM system with a monocular camera mounted onto the gripper of a collaborative robot. We discuss and propose a solution to the pose space conversion problem. Finally, we present several criteria to analyze the 3D reconstruction accuracy. These could be used as guidelines to improve the accuracy of 3D reconstructions with monocular LSD-SLAM and other SLAM based solutions.
We present a new method for detecting gait disorders according to their stadium using cluster methods for sensor data. 21 healthy and 18 Parkinson subjects performed the time up and go test. The time series were segmented into separate steps. For the analysis the horizontal acceleration measured by a mobile sensor system was considered. We used dynamic time warping and hierarchical custering to distinguish the stadiums. A specificity of 92% was achieved.
Fitting 3D Morphable Face Models (3DMM) to a 2D face image allows the separation of face shape from skin texture, as well as correction for face expression. However, the recovered 3D face representation is not readily amenable to processing by convolutional neural networks (CNN). We propose a conformal mapping from a 3D mesh to a 2D image, which makes these machine learning tools accessible by 3D face data. Experiments with a CNN based face recognition system designed using the proposed representation have been carried out to validate the advocated approach. The results obtained on standard benchmarking data sets show its promise.
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.
This paper investigates the evaluation of dense 3D face reconstruction from a single 2D image in the wild. To this end, we organise a competition that provides a new benchmark dataset that contains 2000 2D facial images of 135 subjects as well as their 3D ground truth face scans. In contrast to previous competitions or challenges, the aim of this new benchmark dataset is to evaluate the accuracy of a 3D dense face reconstruction algorithm using real, accurate and high-resolution 3D ground truth face scans. In addition to the dataset, we provide a standard protocol as well as a Python script for the evaluation. Last, we report the results obtained by three state-of-the-art 3D face reconstruction systems on the new benchmark dataset. The competition is organised along with the 2018 13th IEEE Conference on Automatic Face & Gesture Recognition.
This work is a report on practical experiences with the issue of interoperability in German practice management systems (PMSs) from an ongoing clinical trial on teledermatology, the TeleDerm project. A proprietary and established web-platform for store-and-forward telemedicine is integrated with the IT in the GPs’ offices for automatic exchange of basic patient data. Most of the 19 different PMSs included in the study sample lack support of modern health data exchange standards, therefore the relatively old but widely available German health data exchange interface “Gerätedatentransfer” (GDT) is used. Due to the lack of enforcement and regulation of the GDT standard, several obstacles to interoperability are encountered. As a partial, but reusable working solution to cope with these issues, we present a custom middleware which is used in conjunction with GDT. We describe the design, technical implementation and observed hindrances with the existing infrastructure. A discussion on health care interfacing standards and the current state of interoperability in German PMS software is given.
This paper addresses what we call the investment question: under what plausible circumstances, if any, can variable renewable energy (VRE, and solar photovoltaic (PV) in particular) be a good investment? Although VRE has been growing rapidly world-wide, it is generally subsidized. Under what cost and market conditions can solar PV flourish without subsidy? We employ solar insolation and market price data from the U.S. and from Germany to gain insight into the investment question. We find that unsubsidized solar PV is or may soon be a justifiable investment, but that market arrangements may play a crucial role in determining success. We end by sketching a proposal that amounts to a reformed capacity market that would afford participation of solar PV.
This paper presents a digitally controlled boost converter IC for high output voltage and fast transient applications. Thus, it is well applicable in automotive and industrial environments. The 3V-to-6V input voltage, 6.3V output voltage, 1A boost converter IC is fabricated in a 180nm BCD technology. Digital control enables cost savings, advanced control concepts, and it is less parameter sensitive compared to common analog control. A 90 ns latency, 6-bit delay line ADC operates with a window concept, meeting high resolution requirements, e.g. in car battery applications. An output voltage live tracking is included for extending the ADC conversion window. A charge pump DAC provides high resolution, monotonicity, and short 128 ns conversion time. Further, a standard digital PI controller is enhanced by a simple but effective ΔV/Δt-intervention control. It results in 2.8x reduced output voltage deviations in case of load steps, scaling down the output capacitor value by the same factor.
Boost converters suffer from a bandwidth limitation caused by the right-half plane zero (RHPZ), which occurs in the control-to-output transfer function. In contrast, there are many applications that require superior dynamic behavior. Further, size and cost of boost converter systems can be minimized by reduced voltage deviations and fast transient responses in case of large signal load transients. The key idea of the proposed ΔV/Δt-intervention control concept is to adapt the controller output to its new steady state value immediately after a load transient by prediction from known parameters. The concept is implemented in a digital control circuit, consisting of an ASIC in a 110 nm-technology and a Xilinx Spartan-6 field programmable gate array (FPGA). In a boost converter with 3.5V input voltage, 6.3V output voltage, 1.2A load, and 500 kHz switching frequency, the output voltage deviations are 2.8x smaller, scaling down the output capacitor value by the same factor. The recovery times are 2.4x shorter in case of large signal load transients with the proposed concept. The control is widely applicable, as it supports constant switching frequencies and allows for duty cycle and inductor current limitations. It also shows various advantages compared to conventional control and to selected adaptive control concepts.
This paper presents a wide-Vin step-down parallel-resonant converter (PRC), comprising an integrated 5-bit capacitor array and a 300-nH resonant coil, placed in parallel to a conventional buck converter. Soft-switching resonant converters are beneficial for high-Vin multi-MHz converters to reduce dominant switching losses, enabling higher switching frequencies. The output filter inductor is optimized based on an empirical study of available inductors. The study shows that faster switching significantly reduces not only the inductor value but also volume, price, and even the inductor losses. In addition, unlike conventional resonant concepts, soft-switching control as part of the proposed PRC eliminates input voltage-dependent losses over a wide operating range, resulting in 76.3% peak efficiency. At Vin = 48 V, a loss reduction of 35% is achieved compared with the conventional buck converter. Adjusting an integrated capacitor array, and selecting the number of oscillation periods, keeps the switching frequency within a narrow range. This ensures high efficiency across a wide range of Vin = 12–48 V, 100–500-mA load, and 5-V output at up to 25-MHz switching frequency. Thanks to the low output current ripple, the output capacitor can be as small
as 50 nF.
This paper presents a dc–dc converter for integration in the power management unit of an ultra-low power microcontroller. The converter is designed to significantly reduce the wake-up energy and startup delay of the supplied core. The use of a minimized output capacitor is the key factor to save the wake-up energy. The converter is buffered with only 56 nF and guarantees a stable output of 1.2 V with a voltage ripple smaller than 30 mV. The controller of the proposed dc–dc converter is based on a predictive peak current control that allows the system to control the energy transfer at extremely low power consumption. The proposed circuit is implemented in 130 nm CMOS technology with an area of only 0.14 mm². It achieves a high conversion efficiency of 92.1% and a small quiescent current of 440 nA. It operates from 1.8 to 3.3 V with a maximum load of 2.65 mA.
The level shifter and the floating gate supply for high-side transistors are a major challenge in high-voltage DCDC converters. This paper presents a high speed and power-efficient level shifter for voltages of up to 50V, suitable for both PMOS and NMOS power FETs. A switching node falling edge detection allows both, a sensitive and safe signal detection. This enables a robust operation during steep dv / dt transitions and a power consumption as low as 4.1 pJ per switching cycle, which is a reduction of more than 40% compared to prior art. An active clamping circuit prevents common mode displacement currents into the high-side supply. The level shifter is implemented in a 180nm BiCMOS technology. Measurements confirm a 50V 120MHz high-speed operation of the level shifter with a rising / falling propagation delay of 1.45 ns / 1.3 ns, respectively. The dv / dt robustness has been confirmed by measurements for transitions up to 6V/ ns.
A wide-bandwidth galvanically isolated current sensing circuit with an integrated Rogowski coil in 180nm CMOS is presented. Exploiting the high-frequency properties of an optimized on-chip Rogowski coil, currents can be measured up to a bandwidth of 75 MHz. The analog sensor front-end comprises a two-stage integrator, which allows a chopper frequency below signal bandwidth, resulting in 2.2 mVrms output noise. An additional integrated Hall sensor extends the measurement range towards DC.
Due to their superior fast-switching performance, GaN transistors show enormous potential to enable compact power electronics in applications like renewable energy, electrical cars and home appliances by shrinking down the size of passives. However, fast switching poses challenges for the gate driver. Since GaN transistors have a low threshold voltage Vt of ~1V, an unintended driver turn-on can occur in case of a unipolar gate control as shown for a typical half-bridge in Fig. 24.2.1 (top left). This is due to coupling via the gate-drain capacitance (Miller coupling), when the low-side driver turns on, causing a peak current into the gate. This is usually tackled by applying a negative gate voltage to enhance the safety margin towards Vt, resulting in a bipolar gate-driving scheme. In many power-electronics applications GaN transistors operate in reverse conduction, carrying the inductor current during the dead time t, when the high-side and low-side switch are off (as illustrated at a high-side switch in Fig. 24.2.1, bottom left). As there is no real body diode as in silicon devices, the GaN transistor turns on in reverse operation with a voltage drop VF across the drain-source terminals (quasi-body diode behavior). As a negative gate voltage adds to VF, 63% higher reverse-conduction losses were measured for a typical GaN switch in bipolar gate-drive operation. This drawback is addressed by a three-level gate voltage (positive, 0V, negative), which at the same time provides robustness against unintended turn-on similar to the bipolar gate driver, proven in [1] for a discrete driver.
This paper presents a fully integrated gate driver in a 180-nm bipolar CMOS DMOS (BCD) technology with 1.5-A max. gate current, suitable for normally OFF gallium nitride (GaN) power switches, including gate-injection transistors (GIT). Full-bridge driver architecture provides a bipolar and three-level gate drive voltage for a robust and efficient GaN switching. The concept of high voltage energy storing (HVES), which comprises an on-chip resonant LC tank, enables a very area-efficient buffer capacitor integration and superior gatedriving speed. It reduces the component count and the influence of parasitic gate-loop inductance. Theory and calculations confirm the benefits of HVES compared to other capacitor implementation methods. The proposed gate driver delivers a gate charge of up to 11.6 nC, sufficient to drive most types of currently available GaN power transistors. Consequently, HVES enables to utilize the fast switching capabilities of GaN for advanced and compact power electronics.
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.
The digital transformation of our life changes the way we work, learn, communicate, and collaborate. Enterprises are presently transforming their strategy, culture, processes, and their information systems to become digital. The digital transformation deeply disrupts existing enterprises and economies. Digitization fosters the development of IT systems with many rather small and distributed structures, like Internet of Things, Microservices and mobile services. Since years a lot of new business opportunities appear using the potential of services computing, Internet of Things, mobile systems, big data with analytics, cloud computing, collaboration networks, and decision support. Biological metaphors of living and adaptable ecosystems provide the logical foundation for self optimizing and resilient run-time environments for intelligent business services and adaptable distributed information systems with service oriented enterprise architectures. This has a strong impact for architecting digital services and products following both a value-oriented and a service perspective. The change from a closed world modeling world to a more flexible open-world composition and evolution of enterprise architectures defines the moving context for adaptable and high distributed systems, which are essential to enable the digital transformation. The present research paper investigates the evolution of Enterprise Architecture considering new defined value-oriented mappings between digital strategies, digital business models and an improved digital enterprise architecture.
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.
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.
The hotspot detection has received much attention in the recent years due to a substantial mismatch between lithography wavelength and semiconductor technology feature size. This mismatch causes diffraction when transferring the layout from design onto a silicon wafer. As a result, open or short circuits (i.e. lithography hotspots) are more likely to be produced. Additionally, increasing numbers of semiconductors devices on a wafer required more time for the lithography hotspot detection analysis. In this work, we propose a fast and accurate solution based on novel artificial neural network (ANN) architecture for precise lithography hotspot detection using a convolution neural network (CNN) adopting a state of-the-art technique. The experimental results showed that the proposed model gained accuracy improvement over current state-of-theart approaches. The final code has been made publicly available.