Start a new topic

Head movement tracking with YEI

I need to use a YEI sensor to record translation (spatial movement).  I am using the accelerometer but it shows a constant force in the Y value.  I can compensate for this well enough by recording the readout and subtracting it.  Also, Linear Acceleration fixes this.  After that I noticed, looking at the data steam in the Suite, the readings dont seem quite right. After movement stops in one direction the opposite movement is shown as well.  So if I lower the sensor It shows something more like a (jagged) sin wave than a dip.  The attached image shows me moving the YEI sensor down about a foot in 1 second.  Why is there a downward AND upward spike?

png

https://www.youtube.com/watch?v=C7JQ7Rpwn2k This google talk on Sensor Fusion hints that an algorithm is needed to get a correct reading of acceleration/movement.  I was under the impression that the YEI API already had this built in, say with the tss_getCorrectedLinearAccelerationInGlobalSpace .Is this correct or do I still need to implement my own algorithm to determine a clean acceleration?

Your sensor rests, then travels 1 foot downwards and comes to a full stop. Therefore your gesture consists of two phases: linear acceleration, necessary to set the sensor in motion, then linear deceleration, necessary to slow down the sensor to a full stop. The YEI studio graph shows both phases. The first phase is gentle, the second quivery, mimicking car suspension at a harsh brake.

 

I'm not sure if this has been resolved yet, and I apologize for the very long response time. 


I mimicked your test using one of our sensors. The readings may be a little different than yours as I was using a ±24g model. I have attached the images of the data chart of the responses.


In test 1: I held the sensor in my hand, moved down swiftly, and stopped about a foot below its starting point. My results were similar to what you described: A negative acceleration followed by a positive spike.  Image downtest1.


in test 2: I help the sensor in my hand, moved down swiftly, and stopped the sensor on my knee, also about a foot below its starting point. The result was a negative acceleration followed by none-to little positive spike. Image downtest2.


The positive spike in acceleration is caused by the sensor coming to rest in your hand. Your arm does not stop immediately, and causes the positive accel when it bounces as it is coming to a stop.



In regards to sensor fusion: The TSS sensors already fuze the acceleromter, magnetometer, and compass. It can output each component individually in raw, corrected, and normalized formates, but also has on-board filters that use all 3 components to find its orientation. It also has the ability to give its Orientation relative to either Magnetic north or to a custom Tare position, which allows for finding the change in orientation from any angle.

downtest2.png
(20.8 KB)
downtest1.png
(22.9 KB)
Login or Signup to post a comment