Home | english  | Impressum | KIT
Photo von Heinz Wörn

Prof. Dr.-Ing. Dr. h.c. Heinz Wörn

Professor im Ruhestand
Tel.: +49 721 608-44006
Fax: +49 721 608-47141
woernPzr3∂kit edu


Zur Person

Professor Wörn studierte Elektrotechnik an der Universität Stuttgart und promovierte dort am Institut für Werkzeugmaschinen mit seiner Arbeit zu dem Thema "Mehrprozessorsteuerungssystem für Werkzeugmaschinen mit standartisierten Schnittstellen". Im Anschluss arbeitete er bei KUKA Schweißanlagen und Roboter GmbH, wo er eine leitende Stellung in Forschung und Entwicklung inne hatte. Professor Wörn ist ein international anerkannter Experte für Roboter und Automation. Seine Erfahrung umfasst Roboteranwendungen, Robotersteuerungen und Sensoren für Roboter, sowie deren Programmmierung und Simulation. Seit 1997 leitet er das Institut für Prozessrechentechnik, Automation und Robotik der Universität Karlsruhe als Professor für "Komplexe Systeme in Automation und Robotik".

Forschungsgebiete

  • Planung, Programmierung, Steuerung, Diagnose und Sensorsysteme für Industrieroboter
  • Autonome, mobile Roboter, Mikroroboter, Serviceroboter, Teleroboter, Autonome Fahrzeuge
  • Planung und Simulation von Anlagen und Fabriken
  • Roboter- und sensorgestützte Chirurgie
  • Mikromontage
  • Modellierung komplexer Systeme in Produktion und Medizin

Safe Human-Robot Interaction using Novel 3D Sensor

AutorJ. Graf and H. Wörn
Jahr2009
Veröffentlicht inAUTOMATION 2009 (to appear)
KurzfassungThis paper presents research results from the BMBF project Lynkeus (16SV2307). One of the goals in Lynkeus is establishing safe human-robot interaction in realtime using a novel camera system called photonic mixture device (PMD). The PMD sensor provides amplitude and depth images in realtime. The demonstrator cell at our laboratory is actually using industrial robots. In the meantime, the focus on realising such a complex system changed from using purely mathematical rigour models applied to image sequences from the PMD (e.g. dynamic contours combined with optical flow estimation [1], [2] and [3]) to pattern oriented image sequence analysis. This step has several advantages: At first, it is possible to estimate the pose, or to be more precise some important kinematical features, of a human person acting within the robot cell in realtime. Second, it is possible to realise a highly efficient and reliable distance measurement between moving actors and working robots using some kind of line sweep spheres technique applied to both kinematical chains – human and robot. Third, as a consequence of the first two points, such a foregoing enables estimation of risk for the person within the robot cell. Dependant on the risk estimation, which is dependant on special features and will be discussed in detail in the full paper, it is possible to derive actions which influence the control of the robot. A simple and efficient strategy guaranteeing safety is limiting the maximal velocity of the robot. The features, which are used to estimate the risk, are generated from patterns which come from depth images of the PMD. Since the PMD is an optical measurement tool it is natural that the image sequences from such a device are corrupted by noise. Therefore the kalman filter is applied to the estimation process in order to stabilize the measurement of the features. This foregoing is applied to position estimation of the human (localization) and orientation estimations of head, torso and the height of the human. The experimental results in the full paper will show some of the accuracies of the measurement using groundtruth material.
Bibtex@article{ ipr_1170857247, author = "{J. Graf and H. W{{\"o}}rn}", title = "{Safe Human-Robot Interaction using Novel 3D Sensor}", year = "2009", journal = "{AUTOMATION 2009 (to appear)}", }
zurück zur Publikationsübersicht