University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Thorsten Dahmen

Associated Doctoral Student in the PhD program since 01.08.2008.
Doctoral Student in the PhD program from 15.10.2007 to 31.07.2008.

advisors

1. Prof. Dr. Dietmar Saupe
2. Prof. Dr. Oliver Deussen

organisational data

Room: Z 704
Tel.: +49 (0)7531 / 88-2220
E-mail: thorsten.dahmen "at" uni-konstanz.de
picture

project description

Simulation and Optimization of Race-Bike Training on Realistic Tracks

The Powerbike Project contributes to the subdiciplines of acquisition, processing, analysis and optimization of physical and physiological performance data in cycling, computer-aided training control, motion analysis, and information visualization. Its goal is to realistically simulate outdoor riding on existing tracks in the lab, extract the relevant information of a multitude of available measurement data and visualize this information for optimal training control and performance prediction. This thesis focuses on the simulation and optimization of race-biking on realistic tracks. This system shall enable athletes to familiarize themselves with even unknown tracks and optimally prepare them for competition.
We designed a simulator based on a Cyclus2 ergometer (RBM Elektronik), which allows the user to mount his personal bicycle. The resistance control relies on a validated P-v-model that describes the relation between pedaling power P and speed v on a cycling track, whose height profiles was measured by differential GPS augmented by height profile estimations of the model based on calibration rides. Electronic gear levers have been incorporated into the model as well as physical limitations of the eddy current brake have been compensated at best. An algorithm to compensate for the inertia mismatch has been developed, yet needs to be implemented using the RS-232 interface, since the standard TCP/IP interface does not provide the necessary sampling rate. The display of the simulator shows a video playback synchronized with the cyclist's current position on the track together with various course and performance parameters. The accuracy of the system has been validated comparing outdoor measurement, simulator measurements, and model predictions.
The optimal pacing strategy is sought by minimizing the time an athlete needs to complete a course according to the P-v-model subject to calibrated physiological dynamics and constraints that reflect the individual endurance capacity of an athlete. This questions leads to a complex optimal control problem for which robust numerical algorithms must be designed. Direct discretization using pseudo-spectral collocation methods as well as the use of saturation functions and system extensions to transform the constrained optimization problem to an unconstrained one, have proven best performance. In the near future, we plan to design custom ergometer tests to calibrate the physiological models. Furthermore, a model predictive control seems appropriate to turn the current open-loop optimal pacing strategy to a closed-loop feedback control.

publications

The following list of publications covers only those, which are or were published during participation at the Graduiertenkolleg / PhD program.

Articles in Journals

201520142011
2015
2014
2011

Conference Papers

20132012201120092008
2013
2012
2011
2009
2008

curriculum vitae

Since 10/2007 Doctoral Student in the PhD Program, University of Konstanz
06/2007 - 09/2007 Scientific Assistant, Interdisciplinary Center for Scientific Computing, University of Heidelberg
10/2000 - 02/2007 Studies of Electrical Engineering and Information Technology (Diplom), RWTH Aachen
11/2005 - 04/2006 Industrial Internship, Drägerwerk AG, Lübeck
02/2004 - 11/2004 Studies of Physics (PGDipSc), University of Auckland (New Zealand)