mpu6050 DMP

### MPU6050 DMP Configuration and Programming Guide

The MPU-6050 is a popular sensor that integrates a 3-axis gyroscope, a 3-axis accelerometer, and an onboard Digital Motion Processor (DMP). Configuring the DMP allows for advanced motion processing capabilities without overloading the main microcontroller.

#### Initialization of MPU6050 with DMP Support

To initialize the MPU-6050 along with its DMP functionality:

#include "I2Cdev.h"
#include "MPU6050_6Axis_MotionApps20.h"

// Class declaration for MPU6050
MPU6050 mpu;

// Define support variables
uint8_t devStatus;      // Return status after device initialization
uint8_t mpuIntStatus;   // Holds actual interrupt status byte from MPU
uint16_t packetSize;    // Expected DMP packet size (default is 42 bytes)
uint16_t fifoCount;     // Number of bytes waiting in FIFO buffer
uint8_t fifoBuffer[64]; // FIFO storage buffer

void setup() {
  Serial.begin(9600);
  
  // Initialize I2C communication as master
  Wire.begin();
  mpu.initialize();

  // Verify connection to MPU-6050
  if (!mpu.testConnection()) {
    Serial.println("MPU6050 connection failed");
    while (true); // Loop forever if unsuccessful
  } else {
    Serial.println("MPU6050 connected successfully");
  }

  // Load and configure the DMP firmware into memory
  devStatus = mpu.dmpInitialize();

  // Ensure calibration time is adequate
  mpu.CalibrateGyro(5);

  // Supply your own gyro offsets here, scaled for min sensitivity
  mpu.setZAccelOffset(1688);

  // Enable DMP, now using a fixed I2C slave address
  mpu.setDMPEnabled(true);

  // Get expected DMP packet size for later comparison
  packetSize = mpu.dmpGetFIFOPacketSize();
}

This code snippet initializes the MPU-6050 and configures it to use the DMP. The `dmpInitialize` function loads necessary firmware onto the chip which enables complex calculations such as quaternion output directly on-chip rather than offloading this task to another processor[^1].

#### Processing Data from DMP

Once initialized properly, data can be read out via interrupts or polling methods depending upon application requirements. Here’s how one might process incoming packets when utilizing interrupt-driven reading:

void loop() {
  // If we have received new data...
  if (mpuInterrupt == true) {

    // Reset our flag so we don't trigger again immediately.
    mpuInterrupt = false;
    
    // Read the current interrupt status
    mpuIntStatus = mpu.getIntStatus();

    // Check if there's any available data within the FIFO queue
    fifoCount = mpu.getFIFOCount();

    // Handle overflow by resetting both FIFO count & pointer registers
    if ((fifoCount >= 1024) || (mpuIntStatus & 0x10)) { 
      mpu.resetFIFO();      
    }
    
    // Only proceed further if at least enough samples are present
    else if (fifoCount >= packetSize){
        
        // Attempt to fetch all pending records up until maximum capacity limit
        while (fifoCount >= packetSize) {
            mpu.getFIFOBytes(fifoBuffer, packetSize);
            
            // Parse raw values stored inside 'fifoBuffer' array
            mpu.dmpGetQuaternion(&q, fifoBuffer);
            mpu.dmpGetGravity(&gravity, &q);
            mpu.dmpGetYawPitchRoll(ypr, &q, &gravity);
          
           // Print results through serial port interface
           Serial.print("y: ");
           Serial.print(ypr[YAW] * 180/M_PI);
           Serial.print(", p: ");
           Serial.print(ypr[PITCH] * 180/M_PI);
           Serial.print(", r: ");
           Serial.println(ypr[ROLL] * 180/M_PI);
           
          // Update remaining unread entries counter variable
          fifoCount -= packetSize;
       }
       
       // Signal completion of handling routine back to hardware level
       mpuIntStatus &= ~0x02;
    }
  }
}

In this example, once an interrupt occurs indicating fresh information has arrived, relevant details like yaw, pitch, roll angles derived from quaternions get printed out continuously during runtime operations.