1 - 10 of 166
-
-
PublicationIntelligente Bild- und Videoauswertung für die Sicherheit( 2023)
-
PublicationExplaining CNN classifier using association rule mining methods on time-series( 2023)Rule-based approaches are known to be one of the most interpretable prediction models, which means that both domain experts and nondomain users can readily grasp the behavior of the model. So far, the rule mining techniques such as SBRL, Rule Regularization, and Gini Regularization have been proposed on time-series for Multilayer Perceptrons (MLPs). While MLPs perform already quite well on time-series, Convolutional Neural Networks (CNNs) outperform these algorithms in some use cases. Therefore, we adapt the existing three methods to produce an interpretable surrogate model, which imitates the behavior of the CNN classifier on time-series data. This makes the decision process more comprehensible for humans. In the experiments we evaluate three methods on the classification task trained on AIS real-world dataset and compare the performance of each method in terms of metrics such as F1-score, fidelity, and the number of rules in the surrogate list. Besides, we illustrate the impact of support, and bins (number of discrete intervals) parameters on the model's performance along with the strengths and weaknesses of each method.
-
PublicationProduktionsnetzwerke der Automobilindustrie: Ausblick 2030( 2023)Automobilhersteller, Zulieferer aller Größenordnungen und die Logistikunternehmen, auf die diese OEMs und Lieferanten sich verlassen, bewegen sich alle auf eine Ära zu, in der Elektromotoren den Verbrennungsmotor ersetzen und Software ein elementarer Faktor ist. Software wird bzw. ist ein zentraler Differenzierungspunkt, der ein Fahrzeug oder einen Hersteller von einem anderen unterscheidet. Inmitten dieses tiefgreifenden Wandels der Automobilindustrie stehen die Lieferketten und Stoffkreisläufe, deren Transparenz und Resilienz, im Mittelpunkt der Wettbewerbsfähigkeit der Automobilhersteller.
-
-
PublicationRoad vehicle automation and its effects on fatigue, sleep, rest and recuperation( 2023)Assisted and automated driving brings new challenges and opportunities when it comes to driver fatigue. With lower levels of vehicle automation, driver monotony and boredom in combination with demanding attentive monitoring leads to higher levels of fatigue, especially during the night when the sleep pressure is high. With higher levels of vehicle automation, when the driver is not required to continuously monitor the roadway and the automation system, task-related fatigue can be counteracted by engaging in non-driving-related activities. Finally, with the highest levels of vehicle automation, it may even become possible for drivers to sleep while on the move. Aside from making it possible for private car drivers to take strategic naps during a drive, this also opens up the possibility for more flexible, risk management-based, hour of service regulations for professional drivers. This chapter summarises the current state of the art on how assisted and piloted driving affects driver fatigue and how automation may facilitate recovery and recuperation from fatigue while on the move. It also covers how automated functions will impact driver monitoring systems and how new ways of counteracting driver fatigue may arise when automation is available.
-
PublicationAdvancements in Vocational Training through Mobile Assistance Systems( 2023)To ensure a sustainable education for trainees, vocational training has to keep up with the elevated digitalization of manufacturing and work. Complicated technologies used in production systems shift the requirements of job training towards a more tech-centered preparation for future employments. Paired with increased diversity among trainees regarding cultural and educational backgrounds, solutions to accommodate both slower and faster learners are sought after. After an initial exemplification of challenges, that vocational training faces, we discuss the general usage of assistance systems in training applications, in order to better the respective results. For a comparison, we examine state-of-the-art digital learning platforms, i.e. assistance systems, that were developed to enhance vocational training. Typically, these platforms offer more or less individually tailored training plans and implement various technologies and interfaces to improve learning speed or quality. Based on the results of this comparison, we propose a new, mobile and user-adaptive assistance system “XTEND for education”, that involves all stakeholders of the training in the process and uses modern technologies to overcome the challenges of current vocational training. By building on top of node graph-based, dynamically generated assignments, AR-based inclusion of training material and equipment and a robust infrastructure to support it, the XTEND assistance system is shown to be capable of meeting requirements of vocational training in the realm of Industry 4.0.
-
PublicationDeep Learning-Based Action Detection for Continuous Quality Control in Interactive Assistance Systems( 2023)Interactive assistance systems have shown to be useful in various industrial settings, in particular those involving human labor like manual assembly of workpieces. Current systems support workers based on different technologies like projection-based augmented reality, hand or tool tracking or automated inspections using computer vision techniques. While these technologies help to increase product quality significantly, existing solutions are not able to monitor the entire process, which makes it difficult to detect process errors. In this paper, we present a deep-learning based approach for continuous on-the-fly quality control within an interactive assistance system. By using labeled video data of an assembly process, a model can be trained that automatically recognizes and distinguishes single actions and thus control the sequence of subsequent work processes. By integrating the system into the interactive assistance systems, users are made aware on any process errors. Besides presenting the concept and implementation of our deep-learning integration into the assistance system, we describe the created industrial assembly-oriented dataset and present the results from our technical evaluation that shows the potential of applying deep-learning methods into interactive assistance systems.
-
PublicationBehaviour of NiAl air plasma sprayed coatings in chlorine-containing atmospheres( 2023)Chlorine-containing atmospheres are encountered in several industrial processes, leading to severe corrosion phenomena. The application of thermally sprayed protective coatings is one of the solutions to improve the resistance of steels to chlorine corrosion at high temperature and to promote their long-term resistance. Thermodynamic calculations allow the establishment of the stability areas of metals as well as their corresponding oxides and chlorides as a function of the chlorine and oxygen partial pressures of the atmosphere. The coefficient of thermal expansion of a cylindrical bulk NiAl specimen was measured using a L75/1550 dilatometer from Linseis. The diffusion experiments were carried out in a tube furnace made of alumina under argon.
-
PublicationQuanten-Ghost-Imaging mit asynchroner Detektion( 2022)Ein fortschrittlicher Aufbau für Quanten-Ghost-Imaging ermöglicht rauscharme 3-D-Bildgebung von entfernten Objekten. Eine wesentliche technische Neuerung liegt in der asynchronen Detektion durch unabhängige SPAD-Detektoren, wodurch sich der Aufbau entscheidend vereinfachen lässt und sich nun für beliebige Entfernungen eignet. Wie jüngste experimentelle Ergebnisse verdeutlichen, besitzt dieser Aufbau das Potenzial, klassische Systeme in vielen Aspekten zu übertreffen.
