1. Alignment Algorithm for the Vestibular Prosthesis
Graduate Student Team:
Monty Rivers
Adam Schofield
Alexander Trusov
By:
Brian Hensley
Mentors:
Dr. A. Shkel
Ilya Chepurko
Introduce self, mentor

2. Vestibular Prosthesis Overview
There are currently various types of VP
The overall goal of the lab is to create a totally implantable vestibular prosthesis
A vestibular Prosthesis is a prosthesis which helps to restore a persons sense of balance. Some symptoms of a vestibular disorder include Vertigo or dizziness, loss of Balance or spatial orientation, or trouble focusing or tracking movement with your eyes. Every year at least 2 million Americans experience some of theses symptoms resulting in over one billion dollars in medical costs. All of the require a head set or bulky. The microsystems lab wants to create a model that would be 8mm wide, using MEMS technology. Currently I am working on prototypes of the gyroscopes so the code and technical aspects will be taken care of before attempts are made to shrink the size of the device.

3. Motivation
During surgery the implementation of the orientation does not have to be as precise
The Vestibular Prosthesis could shift later depending what it is attached to
Now if we had a prosthesis which was the actual size, some problems we would run into are that the orientation of the gyroscopesSoftware implementation is easier to change than hardware implementation

4. Project Overview Areas which I am involved: Rotational axes algorithm
3D Modeling
The VP is meant to mimic the natural V organs. I worked on issues which will arise when the VP is miniaturized. Not issues on a macro scale. I used a prototype to test the alignment algorithm and computer modeling to test the strength of cube.

5. Goal
Explain the red arrows signify the sensing axis of the gyros and we wish to change these to be in alignment with those of the person

6. The Vestibular Organs
Briefly describe the VO’s. Utricle and Saccule are linear acceleration. Point out the three Semicircular canals. Roughly orthogonal. Point out Cochlear nerve (hearing only).

7. How the Signal Travels Natural System Vestibular Prosthesis Gyroscope
Rotational
Axis
Algorithm
On rotation of the head the endolymph in the semicircular canals flows up against the cupula (which is located inside of the ampulla). The cupula then deflects. Imbedded in the cupula are hair cells. This deflection stimulates them and they pass the signal to your brain which interprets it as rotation. In the VP the Gyroscope is rotated and it outputs a set voltage to a transfer function. This changes the signal and modifies it to a signal which will be interperted by your brain. The rotational axis algorithm modifies this signal and makes it proportional to the angle of incline.
Transfer
Function

8. Transfer Function
The flow rate of endolymph in the ear was characterized in the following equation, by Fernandez and Goldberg in 1971:
The input voltage, from the gyro, becomes the input for this equation as it is transformed to the s and z domain. Time constants are important. Unfortunately, the only ones we have are from monkeys. Further research needs to be done into humans. The time constants effect the asymptotes of the graph.

9. Graph of Model
A few major differences, little bit of noise on graph. And at the ends should go off to sides, but slopes down. This is due to the range of frequencies we used.
Graphic by Ilya Chepurko

10. Rotational Algorithm-Scalar Projection
The first rotational algorithm is based off of vector projection. W is the sensing axis of the gyro and V is the direction it should be sensing so the correct amount it should be relaying is U. By taking the scalar projection of w onto v we only want the magnitude, since the direction is already known.

11. Tests Electrical Simulation Single Gyroscope on rate table Flat
Inclined
The gyroscopes on rate tables will be at inclines so their signal will not be full signal of rotation

12. Electrical Simulation
Wanted to see if the microchip accurately responded to input voltage by producing graph of transfer function
For the experimental set up, microchip w/o gyro, but had wires leading to input on board. Used DC power supply and AC voltage to simulate rotation. Monitored output with dynamic signal analyzer.

13. Results of Electrical Simulation
The graphs line up for the most part, except in two regions. In region one the frequency is small compared to the noise so the electronic noise overpowers the result. In region two the frequency being tested approaches the Nyquist frequency which causes the analog filter in the code to not function properly.
Graphic by Ilya Chepurko

14. Rate Table Test of a Single Gyro
The Flat Tests
For Gyro was put on the rate table with no incline and spun at different frequencies. This was done to ensure The Gyro responded to actual rotation the same it is did to electrical simulation

15. Results of Rotation without Incline
Data points line up very well. Tests were redone and results were improved.

16. Incline Test Did not have a tilt table to test, had to improvise
Used 4 different objects to prop gyro up against giving 4 angles:
8.45
28.38
50.31
59.32
While doing the incline test, was able to apply the rotational algorithm to account for the angle. Wanted to test wide range of angles. Did what was possible

17. Incline Tests
All results close to transfer function. Could only test limited range along TF. (.1Hz to 25Hz) This is an encouraging test which shows that a system of three gyros working together could accurately model motion in three dimensions.

18. Vestibular Prosthesis
Unfolded Box
Folded Box
Another issue with the VP is strength its design. The current design of the VP is a cube. Cube is an ideal shape to use for VP. Need three orthogonal planes for linear accelerators and single axis gyro. Will be on inside of cube
3D Model Designed by Monty Rivers

19. Stress in Latches
On Assembly of Cube Latches break. Needed to do strength analysis tests to see how much displacement they can undergo. Pictures show stress where latch meets with wall.

20. 2-D Silicon Beam Test
The latch can be simplified into a silicon beam in 2D because we only want to measure displacement in 2d and width (into the screen) won’t effect the stresses we are looking at. One end was constrained while the other was displaced. Picture shows beam undergoing displacement and color shows the von Mises stress in beam.

21. Width of Latch
Current width of beam microns. 7 different widths were tested with same displacements to see how they react. The thinner beams deflect more with less stress. This graph is shown with a factor of safety of 2,3, and 4 for the fracture stress of silicon (about 7000 Mpa)

22. Conclusion and Future Work
Either Rotational Algorithm would work
Need to test to see which one would be faster
Apply Rotational Axis Algorithm to Three axis system
Strength of VP

23. Acknowledgements NSF UCI Dr. Shkel and Ilya Chepurko Microsystems Lab
Monty Rivers
Adam Schofield
Alex Trusov
IM-SURE
Said Shokair
IM-SURE Fellows
SAP