|Bibliography||Brucker, Felix: Analysis of Methods for Segmentation and Representation of Time Series for the Recognition of Motion Pattern. |
University of Stuttgart, Faculty of Computer Science, Electrical Engineering, and Information Technology, Bachelor Thesis No. 2350 (2012).
43 pages, english.
|CR-Schema||I.2.1 (Applications and Expert Systems)|
I.2.3 (Deduction and Theorem Proving)
I.2.6 (Artificial Intelligence Learning)
I.2.11 (Distributed Artificial Intelligence)
I.5.1 (Pattern Recognition Models)
I.5.3 (Pattern Recognition Clustering)
In the context of the EU project RoboEarth robots shall exchange their knowledge, in the form of platform independent information, in a worldwide network. The goal is that robots get access to knowledge and experience other robots collected to solve so far unknown problems. To provide complex motion sequences in order to solve such problems, it is necessary to represent them in a platform independent way. Therefore, in a pre-process, the movement records have to be divided into different segments and labelled clearly. In this thesis such a technique for performing this task is analysed and tested. Based on multidimensional continuous sensor signals using Microsofts Kinect, motion sequences are divided into single, recurring movement primitives. For this pur- pose, time series of the positions of different body parts are discretised and represented and later recognised with the help of Hidden Markov Models.
|Full text and|
|PDF (2971857 Bytes)|
|Department(s)||University of Stuttgart, Institute of Parallel and Distributed Systems, Image Understanding|
|Entry date||May 7, 2012|