Student ID, Degree Automatic Control Invariant Representation of Rigid Body Motion Trajectories

Problem description: In robotics, especially cognitive robotics, there is a need to represent human or robot motion in a way that facilitates recognition, classification, and characterization of an executed motion, as well as model based generation of a motion that is adapted to a specific task and task environment. To this end, coordinate-free, scale and view invariant representation of rigid body motion trajectories has been proposed [1], [2], [3]. In this Bachelor Thesis work the student has to implement a novel algorithm to compute an invariant representation starting from motion data (pose and/or twist). To make possible the reproduction of the motion on real robots, the algorithm has to be bi-directional, i.e. invariants can be computed from Cartesian data and vice versa. Tasks: • Literature review on rigid body kinematic and invariant representation of motion trajectories • Invariant representation algorithm implementation (Matlab) • Test the algorithm in real scenario