| |
|
| [ title ] |
 |
HybRat - Brain-Machine Interface for Trajectory Prediction |
[ abstract ]
 |
|
Goal: Implementation of a trainable brain-machine interface for real-time trajectory prediction using semi-invasive surface electrode-arrays and a bio-inspired control architecture for a hybrid rat-machine system.
Results: After 4 1/2 months work (Diploma Thesis), semi-invasive soft multi-array electrodes have been designed for the implantation on the Pia Mater of Wistar rats.
This includes electronics for signal pre-conditioning, a complex structure for chronical fixation on the skull, and the development of suitable implantation procedures (see gallery below).
The specifically developed software in C++ includes modules for the recording of brain signals (16 channels @ 10 kHz), the simultaneous tracking of the rat trajectory with an overhead camera (based on Open Computer Vision Library), and the control of a simulated car in Open Dynamics Engine based on the tracking data.
Videos of the successful tracking of running rats and the trajectory of the simulated car can be found below. Brain signals were recorded shortly before the deadline, however, their significance has yet to be proven.
Closely related work will be continued at the Medical Engineering and Life Science Laboratory, e.g. the RatCar project by O. Fukayama (fellow researcher). |
| [ location ] |
|
University of Tokyo, Graduate School of Information Science and Technology, Departement of Information Science and Technology, Medical Engineering and Life Science Laboratory, Tokyo, Japan |
| [ framework ] |
|
Diploma Thesis for the Swiss Federal Institute of Technology |
| [ duration ] |
|
November 2003 - March 2004 |
|
 |
| [ report ] |
|
Request Diploma Thesis Report (PDF, 23MB, 80 pages) over online form.
|
[ videos ]
|
|
|
Video illustrating the camera tracking of a white Wistar rat running in the test area (trail indicated in blue) followed by a simulated car controlled based on tracking information (simulation in separate window, trail indicated in yellow). The simulation is used to generate data for supervised training while minimizing the Frame-Of-Reference problem. Movie encoding: DivX Codec
|
|
|
|
Video illustrating the camera tracking of a Wistar rat with brain implants during a real recording of brain signals (black cable for recordings visible, movie encoding: DivX Codec).
|
| [ gallery ] |
|
|
|
|
