Kalman filter to estimate altitude and climbrate(sinkrate) by fusing altitude and vertical acceleration data for lag-and-overshoot-free output.
The filter also estimates a third parameter, acceleration bias. In practice, this is the residual bias after the accelerometer sensor is calibrated. It can drift unpredictably with environmental conditions such as temperature and humidity.
The project history is at http://pataga.net/imukalmanvario.html. For algorithm documentation, see "imu_kalman_filter_notes.pdf". imuvario.cpp has example pseudo-code usage of the kalman filter API. This is the main initialization and loop code for an altimeter-vario.
For algorithm comparisons on real data, see http://github.com/har-in-air/ESP32_IMU_BARO_GPS_LOGGER. The /docs directory has a downloaded binary data log file, code used to extract the sensor data from the log, and code to process it. The processing code compares two Kalman filters that output filtered altitude data. The first filter KFZ only uses the pressure sensor derived altitude, and the second filter KFZA fuses altitude and acceleration data for lag-and-overshoot free output. The spreadsheet http://github.com/har-in-air/ESP32_IMU_BARO_GPS_LOGGER/blob/master/docs/FilterResults.ods has the results.
Update: You can find the most recent algorithm documentation and implementation at http://github.com/har-in-air/ESP32_IMU_BARO_GPS_VARIO. This is a GPS/altimeter-vario for recreational gliding.