Need Launch Data to Test EKF Program

[bimmui]

Hello everyone,

I'm currently building a program for a flight computer that can estimate what state the rocket is in. I've been learning about Kalman Filters so I can denoise the data I get from my sensors and wanted to test it out on actual launch data to simulate how well it works and debug the program. Does anyone know where I can go about getting my hands on some data to use? I've been wanting to see how well it works with data from a LSM9DS1 and BMP388 in particular but any data I can use is well worth it.

 

[waltr]

In the ZIP file is Raw data from LSM6DSOX, LIS3MDL and BMP390 sensors on boards from AdaFruit. All units are metric as per the Adafruit Arduno libraries.
These are .csv files which open in Excel
1st column is a time stamp in msec.
Acc (AX,AY,AZ) in m/s/s
Gyro (GX,GY,GZ) in Rads/s
Mag mX,mY,mZ) in uT
Alt (BA) in meters
T is temperature in deg C from LSM6
BT is temperature in deg C from BMP390
BP is baro pressure

One is a single stage and other a 2-stage (airstart).

Once you have actual hardware write code to simply LOG the RAW data. Then you have all the time to work out how a Kalman filter and the other math. I use Excel for quick and dirty then Python to Calibrate the data and extract Flight states.

Once you know what to do with the data then code it into the flight computer.

Have fun

 

[cerving]

Why don't you just fly it and get some data that way?

 

[Spacedog49Krell]

Here are two 250Hz data flights using an MPU-6050 IMU and a BMP280 altimeter. The data format is very similar to waltr's.

If you are using the Kalman algorithm to filter noise, you are using it for the wrong application. Digital sensors have built-in oversampling and filtering routines. There should be no need for further filtering if you set the oversampling and internal filter to your application and sampling speed. The Kalman algorithm predicts where your rocket will be at the next data sampling or some future time.