How to remove the gravity from accelerometer

Other Hardware Sensors
6

Could the roll,pitch, yaw help me to reach my object to have always a right orientation of imu?

/**
   What is this file?
   This is an arduino example that calculates accurate roll,pitch,yaw from raw gyro/accelerometer data
   It has a calibration stage which removes most of the gyro drift and a complementary filter
   that combines gyro and accelerometer angles to produce roll/pitch angles that don't drift (like the gyro angle) and aren't noisy
   (like the accel angles). As there is no magnetic compass on the nano iot 33, it's not possible to 'complement' the yaw
   - hence yaw will drift and is 'relative'.
*/
#include <Arduino_LSM6DS3.h>
#include <Wire.h>

float accelX,            accelY,             accelZ,            // units m/s/s i.e. accelZ if often 9.8 (gravity)
      gyroX,             gyroY,              gyroZ,             // units dps (degrees per second)
      gyroDriftX,        gyroDriftY,         gyroDriftZ,        // units dps
      gyroRoll,          gyroPitch,          gyroYaw,           // units degrees (expect major drift)
      gyroCorrectedRoll, gyroCorrectedPitch, gyroCorrectedYaw,  // units degrees (expect minor drift)
      accRoll,           accPitch,           accYaw,            // units degrees (roll and pitch noisy, yaw not possible)
      complementaryRoll, complementaryPitch, complementaryYaw;  // units degrees (excellent roll, pitch, yaw minor drift)

long lastTime;
long lastInterval;

void setup() {

  Serial.begin(57600);
  pinMode(LED_BUILTIN, OUTPUT);

  // this sketch will wait until something connects to serial!
  // this could be 'serial monitor', 'serial plotter' or 'processing.org P3D client' (see ./processing/RollPitchYaw3d.pde file)
  while (!Serial);

  if (!IMU.begin()) {
    Serial.println("Failed to initialize IMU!");
    while (1);
  }

  calibrateIMU(250, 250);

  lastTime = micros();

}

/*
  the gyro's x,y,z values drift by a steady amount. if we measure this when arduino is still
  we can correct the drift when doing real measurements later
*/
void calibrateIMU(int delayMillis, int calibrationMillis) {

  int calibrationCount = 0;

  delay(delayMillis); // to avoid shakes after pressing reset button

  float sumX, sumY, sumZ;
  int startTime = millis();
  while (millis() < startTime + calibrationMillis) {
    if (readIMU()) {
      // in an ideal world gyroX/Y/Z == 0, anything higher or lower represents drift
      sumX += gyroX;
      sumY += gyroY;
      sumZ += gyroZ;

      calibrationCount++;
    }
  }

  if (calibrationCount == 0) {
    Serial.println("Failed to calibrate");
  }

  gyroDriftX = sumX / calibrationCount;
  gyroDriftY = sumY / calibrationCount;
  gyroDriftZ = sumZ / calibrationCount;

}

/**
   Read accel and gyro data.
   returns true if value is 'new' and false if IMU is returning old cached data
*/
bool readIMU() {
  if (IMU.accelerationAvailable() && IMU.gyroscopeAvailable() ) {
    IMU.readAcceleration(accelX, accelY, accelZ);
    IMU.readGyroscope(gyroX, gyroY, gyroZ);
    return true;
  }
  return false;
}

// the loop function runs over and over again forever
void loop() {

  if (readIMU()) {
    long currentTime = micros();
    lastInterval = currentTime - lastTime; // expecting this to be ~104Hz +- 4%
    lastTime = currentTime;

    doCalculations();
    printCalculations();

  }

}

/**
   I'm expecting, over time, the Arduino_LSM6DS3.h will add functions to do most of this,
   but as of 1.0.0 this was missing.
*/
void doCalculations() {
  accRoll = atan2(accelY, accelZ) * 180 / M_PI;
  accPitch = atan2(-accelX, sqrt(accelY * accelY + accelZ * accelZ)) * 180 / M_PI;

  float lastFrequency = 1000000 / lastInterval;
  gyroRoll = gyroRoll + (gyroX / lastFrequency);
  gyroPitch = gyroPitch + (gyroY / lastFrequency);
  gyroYaw = gyroYaw + (gyroZ / lastFrequency);

  gyroCorrectedRoll = gyroCorrectedRoll + ((gyroX - gyroDriftX) / lastFrequency);
  gyroCorrectedPitch = gyroCorrectedPitch + ((gyroY - gyroDriftY) / lastFrequency);
  gyroCorrectedYaw = gyroCorrectedYaw + ((gyroZ - gyroDriftZ) / lastFrequency);

  complementaryRoll = complementaryRoll + ((gyroX - gyroDriftX) / lastFrequency);
  complementaryPitch = complementaryPitch + ((gyroY - gyroDriftY) / lastFrequency);
  complementaryYaw = complementaryYaw + ((gyroZ - gyroDriftZ) / lastFrequency);

  complementaryRoll = 0.98 * complementaryRoll + 0.02 * accRoll;
  complementaryPitch = 0.98 * complementaryPitch + 0.02 * accPitch;
}

/**
   This comma separated format is best 'viewed' using 'serial plotter' or processing.org client (see ./processing/RollPitchYaw3d.pde example)
*/
void printCalculations() {
  Serial.print(gyroRoll);
  Serial.print(',');
  Serial.print(gyroPitch);
  Serial.print(',');
  Serial.print(gyroYaw);
  Serial.print(',');
  Serial.print(gyroCorrectedRoll);
  Serial.print(',');
  Serial.print(gyroCorrectedPitch);
  Serial.print(',');
  Serial.print(gyroCorrectedYaw);
  Serial.print(',');
  Serial.print(accRoll);
  Serial.print(',');
  Serial.print(accPitch);
  Serial.print(',');
  Serial.print(accYaw);
  Serial.print(',');
  Serial.print(complementaryRoll);
  Serial.print(',');
  Serial.print(complementaryPitch);
  Serial.print(',');
  Serial.print(complementaryYaw);
  Serial.println("");
}

This is found in github but I have no a good idea about the plot