C04: Learning of Task Models for Manipulation Tasks in the Circular Factory
The goal is to learn task models for bimanual manipulation tasks from human observation and to adapt and generalize them. A library of task models will be created, which will be continuously extended and improved based on the execution feedback humanoid and two-armed robots in the learning cell. It deals with the segmentation of human demonstrations. Methods for learning task models from human demonstration will be developed. It will adapt and generalize learned task models to similar products, while will address the mapping of task models to robots. It is dedicated to data collection and annotation.
Principal Investigators
Prof. Asfour has many years of experience and expertise in the field of humanoid robotics with a focus on holistic software-hardware robot architectures, grasping and manipulation, and imitation learning