Master Projects 2012

Janusz Michalik – Computational fluid dynamics of turbulent blood flow

Supervisor: Dr. Simone Deparis

Co-supervisor: Gwenol Grandperrin

Blood flow can be either laminar or turbulent. In this work, we use the DNS implementations of LifeV to try to capture turbulent characteristics in a stenosis and brachiocephalic artery to help determine if turbulence models should be implemented.
Along the way, we explain the basics of turbulence, methods for quantifying turbulent flow, and assumptions that are needed for the simulation of blood flow.
This is followed by an explanation of how the Navier-Stokes equations are solved and a brief introduction to turbulence modeling.
Lastly, the simulated cases are explained in detail and their corresponding preliminary results.

Rui Wang – Development of measurement uncertainty calculator based on Monte Carlo technique

EPFL supervisor: Prof. Fabio Nobile

   Federal Office of Metrology (METAS), Bern, Switzerland
   Responsibles: Dr. Markus Zeier, Michael Wollensack
Implementation into Metas.UncLib.
The project would include the following tasks:
1) EVALUATION: pseudo random number generators, generation of random variates for different distributions, sampling techniques
2) DEVELOPMENT: Implementation strategy, propagation of distributions, storage concept
3) IMPLEMENTATION: Implementation in C#, performance tests and validation


Hainan Hu – Analysis and synthesis of conversational nonverbal behavior in human-robot interactions

EPFL supervisor: Jean-Marc ODOBEZ

    About Idiap Research Institute, Martigny, Switzerland

Responsible: Dr. A. Popescu-Belis

The goal of this master project is to contribute towards endowing a humanoid robot with the capacity to interact with multiple people at the same time. This project has two tracks. In the first track, we investigate various methods to detect a speaker and the addressee on a recorded HRI data corpus. In the second track, we then implement a real-time method on the robot to determine the speaker and use this influence the looking behavior of the humanoid robot. We performed experiments to validate our approach. The master project contributes to the European HUMAVIPS project.