A5082150620- Robot Manipulation and Learning
- Human Pose and Action Recognition
- Psychology, Coaching, and Therapy
- AI-based Problem Solving and Planning
- Robotic Mechanisms and Dynamics
- Anomaly Detection Techniques and Applications
- Gaze Tracking and Assistive Technology
- Psychoanalysis and Social Critique
- Hand Gesture Recognition Systems
- Public Administration and Political Analysis
- German Literature and Culture Studies
- Manufacturing Process and Optimization
- Robotics and Automated Systems
- Health Sciences Research and Education
- Flexible and Reconfigurable Manufacturing Systems
- Modular Robots and Swarm Intelligence
- Health Policy Implementation Science
- Corporate Social Responsibility and Sustainability
- Nursing Diagnosis and Documentation
- Autonomous Vehicle Technology and Safety
- Corporate Management and Leadership
- Transportation and Mobility Innovations
- Human Motion and Animation
- Teleoperation and Haptic Systems
- Industrial Vision Systems and Defect Detection
Karlsruhe Institute of Technology
2017-2024
European University of Applied Sciences
2024
CE Technologies (United Kingdom)
2023
Japan Science and Technology Agency
2022
Naver (South Korea)
2022
Walter de Gruyter (Germany)
2022
Ingenieurgesellschaft Auto und Verkehr (Germany)
2019
Recognizing human actions is a vital task for humanoid robot, especially in domains like programming by demonstration. Previous approaches on action recognition primarily focused the overall prevalent being executed, but we argue that bimanual motion cannot always be described sufficiently with single label. We present system framewise classification and segmentation demonstrations. The extracts symbolic spatial object relations from raw RGB-D video data captured robot's point of view order...
ABSTRACT Introduction: Evidence-based practice (EBP) is an important component of clinical in public health. Its implementation involves interpreting scientific studies and then applying this knowledge to decision-making. In Germany, the therapy professions are often trained non-academic medical schools, only a small number therapists university graduates. Aims: This study assessed current status EBP among physiotherapists, occupational therapists, speech language determine whether...
We introduce a multi-functional robotic gripper equipped with set of actions required for disassembly electromechanical devices. The consists robot arm 5 degrees freedom (DoF) manipulation and jaw 1-DoF rotation joint closing joint. system enables in 7 DoF offers the ability to reposition objects hand perform tasks that usually require bimanual systems. sensor includes relative absolute encoders, force pressure sensors provide feedback about interaction forces, tool- mounted camera screw...
Learning temporal relations between actions in a bimanual manipulation task is important for capturing the constraints of required to achieve task's goal. However, given several demonstrations task, problem identifying true dependencies - if there are any very challenging due contradictions. We propose model-driven approach learning models from multiple human that represents on two levels. First, sets exhibit tight coupling, and second, these actions. build Allen's interval algebra as...
Zusammenfassung Agile Produktionssysteme vereinen ein hohes Maß an Flexibilität und Wandlungsfähigkeit. Diese Qualitäten sind insbesondere in einer Umgebung mit hoher Unsicherheit entscheidend, beispielsweise im Kontext von Remanufacturing. Remanufacturing beschreibt den industriellen Prozess der Aufbereitung Gebrauchtteilen, sodass diese vergleichbare technische Eigenschaften wie Neuteile zurückerlangen. Aufgrund Ressourcenknappheit regulatorischer Vorgaben nimmt die Bedeutung zu. Bedingt...
Zusammenfassung Der Mensch ist die flexibelste, aber auch eine teure Ressource in einem Produktionssystem. Im Kontext des Remanufacturings sind Roboter kostengünstige Alternative, jedoch deren Programmierung oft nicht rentabel. Das Programmieren durch Vormachen verspricht flexible und intuitive selbst von Laien durchführbar wäre, doch hierfür zunächst Erfassung Interpretation Handlungen Menschen nötig. Diese Arbeit stellt multisensorielle, robotergestützte Plattform vor, welche zweihändiger...
Zusammenfassung Hintergrund Die wissenschaftliche Kompetenz bei der Anwendung Evidenzbasierter Praxis (EBP) in den Berufsgruppen Physiotherapie (PT), Ergotherapie (ET) und Sprachtherapie (ST) variiert stark, aufgrund derzeitigen divergenten Ausbildungsstruktur (Ausbildung/Studium) Deutschland. Ziel qualitative Studie evaluierte mithilfe von Expert*inneninterviews die Umsetzungsbarrieren vorangegangenen EBP-Studie identifizierte Expert*innenmeinungen zu möglichen Lösungsansätzen für...
Learning task models of bimanual manipulation from human demonstration and their execution on a robot should take temporal constraints between actions into account. This includes (i) the symbolic level such as precedence relations or overlap in execution, (ii) subsymbolic duration different actions, starting end points time. Such are crucial for planning, reasoning, exact timing robot. In our previous work, we addressed learning demonstrated how can leverage this knowledge to respond...
Abstract Process automation is essential to establish an economically viable circular factory in high-wage locations. This involves using autonomous production technologies, such as robots, disassemble, reprocess, and reassemble used products with unknown conditions into the original or a new generation of products. complex highly dynamic issue that high degree uncertainty. To adapt robots these conditions, learning from humans necessary. Humans are most flexible resource they can their...
In this paper, we present a novel approach for learning bimanual manipulation actions from human demonstration by extracting spatial constraints between affordance regions, termed constraints, of the objects involved. Affordance regions are defined as object parts that provide interaction possibilities to an agent. For example, bottom bottle affords be placed on surface, while its spout contained liquid poured. We propose learn changes in construct action models representing interactions. To...
In this paper, we present a novel approach for learning bimanual manipulation actions from human demonstration by extracting spatial constraints between affordance regions, termed constraints, of the objects involved. Affordance regions are defined as object parts that provide interaction possibilities to an agent. For example, bottom bottle affords be placed on surface, while its spout contained liquid poured. We propose learn changes in construct action models representing interactions. To...
There have been many proposals for algorithms segmenting human whole-body motion in the literature. However, wide range of use cases, datasets, and quality measures that were used evaluation render comparison challenging. In this paper, we introduce a framework puts segmentation on unified testing ground provides possibility to allow comparing them. The features both set known from literature novel approach tailored algorithms, termed Integrated Kernel approach. Datasets recordings, provided...
Cognitive agents such as humans and robots perceive their environment through an abundance of sensors producing streams data that need to be processed generate intelligent behavior. A key question cognition-enabled AI-driven robotics is how organize manage knowledge efficiently in a cognitive robot control architecture. We argue, memory central active component architectures mediates between semantic sensorimotor representations, orchestrates the flow events different processes provides...
Movement primitives (MPs) are compact representations of robot skills that can be learned from demonstrations and combined into complex behaviors. However, merely equipping robots with a fixed set innate MPs is insufficient to deploy them in dynamic unpredictable environments. Instead, the full potential remains attained via adaptable, large-scale MP libraries. In this paper, we propose seven fundamental operations incrementally learn, improve, re-organize To showcase their applicability,...
Recognizing human actions is a vital task for humanoid robot, especially in domains like programming by demonstration. Previous approaches on action recognition primarily focused the overall prevalent being executed, but we argue that bimanual motion cannot always be described sufficiently with single label. We present system frame-wise classification and segmentation demonstrations. The extracts symbolic spatial object relations from raw RGB-D video data captured robot's point of view order...