Learning Dynamic Systems for Intention Recognition in Human-Robot-Cooperation
Peter Krauthausen · Repository KITopen (Karlsruhe Institute of Technology) · 2012
This thesis is concerned with intention recognition for non-verbal human-robot-cooperation. The problems of intention recognition based on uncertain and incomplete observations with detailed models in real-time are addressed by a consistent uncertainty processing, automatic model identification of the employed nonlinear stochastic dependencies and situation-specific inference in large dynamic Bayesian networks.