Difference between revisions of "User Manual/4. Sensor calibration"
|Line 65:||Line 65:|
*Go to the next face, and repeat these steps.
*Go to the next face, and repeat these steps.
Once you are done with the last face, place the box on the table and press a key. The program will take some seconds to process the information, just wait and press
Once you are done with the last face, place the box on the table and press a key. The program will take some seconds to process the information, just wait and press when you are asked to.
That’s it, now you have a calibrated and much more accurate sensor. '''You can keep using it like this for some time'''.
That’s it, now you have a calibrated and much more accurate sensor. '''You can keep using it like this for some time''' .
Revision as of 11:27, 9 April 2019
This whole system is based on an amazing technology called MEMS. It stands for Micro Electro Mechanical Systems. What it means is that inside a chip that is usually smaller than 1mm2 the manufacturers are able to squash not only the logic ICs, but also microscopic moving parts. Otherwise they wouldn’t be able to detect physical properties like acceleration, wouldn’t they?
Inertial sensors based on this technology are inexpensive, and they deliver amazing results, but they have the drawback of having some inherent noise and offset on the delivered values. In addition, magnetometers are easily disturbed by the presence of metallic masses on the surroundings generating large distortions of the perceived magnetic field.
The good news is that most of these problems can be resolved with a proper calibration. On top of this, all that it takes to do it is your hands, some patience and a special cardboard box that you can download from our website. This calibration should only be done before the first time you use the system, or when you move to a drastically different place in terms of global latitude and electromagnetic distortion.
Build the cardboard calibration box
Start by downloading the printable unfolded calibration box, print it in A3 paper, glue it to some piece of cardboard, cut it following the solid lines, fold it following the dotted lines. The result should be something like this:
Preparing the hardware
In order to perform the calibration, you will need the hub and SBC to be connected and working (as described in §2 of this manual).
You also need to have SSH access to the SBC (as described in §2.1.5 of this manual).
Insert one K-Ceptor in the socket formed by the cardboard on the inside, make sure the K-Ceptor stays fixed in a stable position' and plug the K-Ceptor to the Hub (using the Gate #1, labeled J1)
Preparing the software
Chordata.xml file to read just one K-Ceptor with the correct translation value
<Armature> <Mux Name="main" id="0"> 0x77 <Branch Name="left" id="1"> CH_1 <!-- This is the gate where to plug your K-Ceptor --> <K_Ceptor Name="Unico" id="2"> 1 <!-- This should be the value of your K-Ceptor --> </K_Ceptor> </Branch> </Mux> </Armature>
Now you can run the notochord, using the
cd bin ./notochord -r <YOUR COMPUTER'S IP ADDRESS>
You will be asked to enter the value and name of the node, and then guided through the steps below.
Buildings have often a metallic structure. If you can try to do it in a place without metallic structure, on the outside, or at least in the most peripheric part of the building (for example: basements are usually not the best places to do it), that will surely lead to better results.
You will be asked to place the calibration box on flat surface facing upwards. This means that the big number 1 should be upwards. Be sure to let the box completely still (sometimes the tension of the wire might move it a little) and press a key.
Then you will be asked to perform the magnetometer calibration procedure. This procedure consists of grabbing the box with your hands and repeating the following steps for each face:
- Hold the box in front of you and make sure to keep the current face as parallel to the ground as possible.
- Put the letter “a” of the current face on the same direction you are looking to.
- While holding the box parallel to the ground, turn it 90º in order to put the letter “b” on the direction you are looking to. Repeat for all 4 letters of the face.
- Starting with the letter "a" in front of you, make some additional random movements. A good technique is drawing some "8" in mid-air.
- Go to the next face, and repeat these steps.
Once you are done with the last face, place the box on the table and press a key. The program will take some seconds to process the information, just wait and press a key when you are asked to. That’s it, now you have a calibrated and much more accurate sensor. You can keep using it like this for some time (the calibration data will be stored on the K-Ceptor's EEPROM memory.
Test the calibration
cd bin ./notochord <YOUR COMPUTER'S IP ADDRESS>
Run the notochord without the
On your computer load the testcube.blend file, and hit the [Receive] button. Rotate your sensor around: the virtual K-Ceptor should reflect the movements of the physical one.
After some random rotations, it is advisable to turn it 90° on some axis and hold it there for a while. Pay attention to any possible drifts of the virtual K-Ceptor from the real position. That would mean that the calibration should be repeated.