Master thesis: Contact estimation

Master thesis:
Contact estimation
For body motion capturing, a popular solution is using inertial measurement units (IMUs). These small
sensors provide among others 3D linear acceleration including gravity and 3D rotational velocity. By
placing them on every major limb segment and taking advantage of a biomechanical model, intrinsic
motion capturing can be performed. A major drawback, however, is the relative nature of these sensors
which makes extrinsic tracking impossible in first place. Double integrating acceleration over time
results quickly in huge errors. A robust contact estimation can bound these errors by giving
information, which point has zero velocity. The goal of this thesis is a robust classification algorithm
based on raw sensor data.
Suggested task:
• Research and evaluate existing approaches to detect zero velocity from inertial measurements.
• Develop and implement a classification algorithm based on a sliding window of raw sensor data.
• Evaluate your approach against the existing algorithms.
Requirements:
• Advanced C++ skills
• Preferably experience with intertial sensors
References:
• Skog, Isaac et al., 2010, Zero-Velocity Detection – An Algorithm Evaluation
This thesis will be performed in the new research group wearHEALTH (Department of Computer
Science, Technical University Kaiserslautern) in cooperation with the Augmented Vision Department
(AV) at DFKI. You will be supervised by a postdoc and a Ph.D. Student from wearHEALTH.
Please send your application to the below contact!
Markus Miezal
Technical University Kaiserslautern
Department for Computer Science
AG wearHEALTH
Gottlieb-Daimler-Straße 48
D-67663 Kaiserslautern
Phone: +49(0)631 205 75-2644
Email: miezal@cs.uni-kl.de
www.wearhealth.org
27.04.2015