Tim Lueth's Research Page

Research in Medical Robotics, AI, and Automated Design

Prof. Tim Lueth began his scientific career after completing his studies in 1989, focusing on neural networks, signal theory, language compression , and supervised learning. During his doctoral research (1989–1993), he specialized in Computer-Aided Manufacturing (CAM), followed by postdoctoral work (1993–1997) on autonomous mobile assembly and manipulation robots within the DFG Collaborative Research Center 314 “Artificial Intelligence and Robotics.”

Since 1997, as Professor of Medical Robotics and Navigation at Charité and Department Director at the Fraunhofer Institute IPK, Lueth pioneered the concept of hands-on robotic systems for surgical environments. His team developed the first certified tactile surgical robots for cranial operations (2000). The first commercially successful dental navigation system, RoboDent, was created in 2001 for Straumann AG. In 2003, his group realized the first 3D C-arm system (Vario 3D) for Ziehm Imaging.

Key collaborators during this period included Andreas Hein, Stefan Weber, Olaf Schermeier, and Ralf Tita. Clinical partners were Jürgen Bier, Martin Klein, Frank Hölzle, Ralf Sodian, Norbert Hosten, Detlef Hildebrand, and Gero Strauss.

Since 2005, Lueth has been Professor in the Department of Mechanical Engineering. In 2006, his group developed the first ENT-dental navigation system for Karl Storz GmbH. In 2013, his research influenced developments later integrated into MAKO (Stryker).

Current Research Focus

• Generative AI for automated design, documentation, manufacturing, and certification of robotic systems
• Development of the SGLib design library for MATLAB (since 2012)
• Integration of MATLAB interfaces between SGLib and OpenAI services via API
• Automation of scientific workflows using AI
• Promotion of young researchers and encouragement of academic entrepreneurship

Behavior-Based Robot Control for Assembly KACORS (1995)

This study by Jochen Heinzmann, Ulrich Rembold, Thomas Längle, and Tim Lueth demonstrated that behavior-based robots can perform complex assembly tasks such as the Cranfield benchmark. Tim Lueth led the development of dynamic task mapping and distributed cooperation protocols, enabling real-time behavior switching. Heinzmann, Rembold, and Längle contributed to system integration and experimental validation. The team extended Francesco Mondadas Khepera real-time OS to support concurrent behaviors. Their work built on foundational concepts from Rodney Brooks and influenced later research by Alexander Zelinsky and others in behavior-based robotics.

Video 1995: KACORS - Khepera based micro robot for assembly tasks (6 Minutes).

Collision-Free Robot Cell Layout Planning (1992)

The ESPRIT-2 CIM-PLATO project (1992) involved Klaus Hoermann, Bernd Welz, Tim Lueth, Jianwei Zhang, and Ulrich Rembold. Tim Lueth achieved a key contribution by developing a method for planning robot production cell layouts enabling collision-free motion execution. Klaus Hoermann transformed potential obstacles into the Cartesian configuration space, while Bernd Welz, Jianwei Zhang, and Ulrich Rembold contributed to system integration and validation. Using Peter Adolphs real-time motion planning library under Henning Tolle, the team demonstrated that geometric motion planning could be achieved without simulation.

References

Lueth, T.C. (1992) Automated planning of robot workcell layouts. Proc. IEEE ICRA, pp.11031108.

Näger, G. & Lueth, T.C. (1994) Development of an Integrated Domain Representation for Configuration, Action Planning, and Layout Planning of Assembly Systems. ISPE/IFAC Conf., pp.8792.

Lueth, T., Osmers, U., Spath, D. & Rembold, U. (1996) Information Infrastructure and Information Processing in Distributed Controlled Production Cells (Informationsinfrastruktur und Informationsverarbeitung in dezentral gesteuerten Produktionszellen). FAIM96 Conf., Atlanta.

Lozano-Pérez, T. (1983) Spatial planning: A configuration space approach. IEEE Trans. Comput., 32(2), 108120.

Latombe, J.-C. (1991) Robot Motion Planning. Kluwer Academic Publishers.

Neural Network Control of a Table Tennis Robot (1990)

This early research (19891991) at TU Darmstadt involved Tim Lueth, Wolfgang Hilberg, Robert Piloty, Kurt Siedenburg, Joachim Wietzke, Rüdiger Bien, Joachim Schürmann, and Werner von Seelen. The robot, primarily constructed by Siedenburg and Wietzke under Hilberg and Piloty, was designed to play table tennis. Tim Lueths main contribution was the development of a neural network system that predicted the balls impact point from dual PSD sensor data, enabling real-time control. Schürmanns work on polynomial classifiers and von Seelens lectures on neural computation strongly influenced Lueths approach.

References

Lueth, T., Wietzke, J., & Bien, R. (1991). Neuronale Netze spielen Tischtennis (Neural Networks Play Table Tennis). In Mustererkennung 1991, 13. DAGM-Symposium (pp. 8287). Springer-Verlag.

Rumelhart, D.E., Hinton, G.E., & Williams, R.J. (1986). Learning representations by back-propagating errors. Nature, 323, 533536.

Kawato, M., Maeda, Y., Uno, Y., & Suzuki, R. (1990). Trajectory formation of arm movement by cascade neural network model based on minimum torque-change criterion. Biological Cybernetics, 62, 275288.

Hopfield, J.J. (1982). Neural networks and physical systems with emergent collective computational abilities. PNAS, 79(8), 25542558.