Suchakhree Srisukprom
/
Test_Megre
k
MPU9250.h
- Committer:
- Suchakhree
- Date:
- 2015-12-09
- Revision:
- 1:426fbd0d126a
- Parent:
- 0:238df339023b
File content as of revision 1:426fbd0d126a:
/***** Library based on MPU-9250_Snowda library. It has been modified by Josué Olmeda Castelló for imasD Tecnología. It uses the mbed I2C class for comunications between the sensor and the master controller. Methods related with data filtering have not been tested. AD0 should be connected to GND. 04/05/2015 *****/ #ifndef MPU9250_H #define MPU9250_H #include "mbed.h" #include "math.h" #define M_PI 3.14159265358979323846 // See also MPU-9250 Register Map and Descriptions, Revision 4.0, RM-MPU-9250A-00, Rev. 1.4, 9/9/2013 for registers not listed in // above document; the MPU9250 and MPU9150 are virtually identical but the latter has a different register map // //Magnetometer Registers #define AK8963_ADDRESS 0x0C<<1 #define WHO_AM_I_AK8963 0x00 // should return 0x48 #define INFO 0x01 #define AK8963_ST1 0x02 // data ready status bit 0 #define AK8963_XOUT_L 0x03 // data #define AK8963_XOUT_H 0x04 #define AK8963_YOUT_L 0x05 #define AK8963_YOUT_H 0x06 #define AK8963_ZOUT_L 0x07 #define AK8963_ZOUT_H 0x08 #define AK8963_ST2 0x09 // Data overflow bit 3 and data read error status bit 2 #define AK8963_CNTL 0x0A // Power down (0000), single-measurement (0001), self-test (1000) and Fuse ROM (1111) modes on bits 3:0 #define AK8963_ASTC 0x0C // Self test control #define AK8963_I2CDIS 0x0F // I2C disable #define AK8963_ASAX 0x10 // Fuse ROM x-axis sensitivity adjustment value #define AK8963_ASAY 0x11 // Fuse ROM y-axis sensitivity adjustment value #define AK8963_ASAZ 0x12 // Fuse ROM z-axis sensitivity adjustment value #define SELF_TEST_X_GYRO 0x00 #define SELF_TEST_Y_GYRO 0x01 #define SELF_TEST_Z_GYRO 0x02 /*#define X_FINE_GAIN 0x03 // [7:0] fine gain #define Y_FINE_GAIN 0x04 #define Z_FINE_GAIN 0x05 #define XA_OFFSET_H 0x06 // User-defined trim values for accelerometer #define XA_OFFSET_L_TC 0x07 #define YA_OFFSET_H 0x08 #define YA_OFFSET_L_TC 0x09 #define ZA_OFFSET_H 0x0A #define ZA_OFFSET_L_TC 0x0B */ #define SELF_TEST_X_ACCEL 0x0D #define SELF_TEST_Y_ACCEL 0x0E #define SELF_TEST_Z_ACCEL 0x0F #define SELF_TEST_A 0x10 #define XG_OFFSET_H 0x13 // User-defined trim values for gyroscope #define XG_OFFSET_L 0x14 #define YG_OFFSET_H 0x15 #define YG_OFFSET_L 0x16 #define ZG_OFFSET_H 0x17 #define ZG_OFFSET_L 0x18 #define SMPLRT_DIV 0x19 #define CONFIG 0x1A #define GYRO_CONFIG 0x1B #define ACCEL_CONFIG 0x1C #define ACCEL_CONFIG2 0x1D #define LP_ACCEL_ODR 0x1E #define WOM_THR 0x1F #define MOT_DUR 0x20 // Duration counter threshold for motion interrupt generation, 1 kHz rate, LSB = 1 ms #define ZMOT_THR 0x21 // Zero-motion detection threshold bits [7:0] #define ZRMOT_DUR 0x22 // Duration counter threshold for zero motion interrupt generation, 16 Hz rate, LSB = 64 ms #define FIFO_EN 0x23 #define I2C_MST_CTRL 0x24 #define I2C_SLV0_ADDR 0x25 #define I2C_SLV0_REG 0x26 #define I2C_SLV0_CTRL 0x27 #define I2C_SLV1_ADDR 0x28 #define I2C_SLV1_REG 0x29 #define I2C_SLV1_CTRL 0x2A #define I2C_SLV2_ADDR 0x2B #define I2C_SLV2_REG 0x2C #define I2C_SLV2_CTRL 0x2D #define I2C_SLV3_ADDR 0x2E #define I2C_SLV3_REG 0x2F #define I2C_SLV3_CTRL 0x30 #define I2C_SLV4_ADDR 0x31 #define I2C_SLV4_REG 0x32 #define I2C_SLV4_DO 0x33 #define I2C_SLV4_CTRL 0x34 #define I2C_SLV4_DI 0x35 #define I2C_MST_STATUS 0x36 #define INT_PIN_CFG 0x37 #define INT_ENABLE 0x38 #define DMP_INT_STATUS 0x39 // Check DMP interrupt #define INT_STATUS 0x3A #define ACCEL_XOUT_H 0x3B #define ACCEL_XOUT_L 0x3C #define ACCEL_YOUT_H 0x3D #define ACCEL_YOUT_L 0x3E #define ACCEL_ZOUT_H 0x3F #define ACCEL_ZOUT_L 0x40 #define TEMP_OUT_H 0x41 #define TEMP_OUT_L 0x42 #define GYRO_XOUT_H 0x43 #define GYRO_XOUT_L 0x44 #define GYRO_YOUT_H 0x45 #define GYRO_YOUT_L 0x46 #define GYRO_ZOUT_H 0x47 #define GYRO_ZOUT_L 0x48 #define EXT_SENS_DATA_00 0x49 #define EXT_SENS_DATA_01 0x4A #define EXT_SENS_DATA_02 0x4B #define EXT_SENS_DATA_03 0x4C #define EXT_SENS_DATA_04 0x4D #define EXT_SENS_DATA_05 0x4E #define EXT_SENS_DATA_06 0x4F #define EXT_SENS_DATA_07 0x50 #define EXT_SENS_DATA_08 0x51 #define EXT_SENS_DATA_09 0x52 #define EXT_SENS_DATA_10 0x53 #define EXT_SENS_DATA_11 0x54 #define EXT_SENS_DATA_12 0x55 #define EXT_SENS_DATA_13 0x56 #define EXT_SENS_DATA_14 0x57 #define EXT_SENS_DATA_15 0x58 #define EXT_SENS_DATA_16 0x59 #define EXT_SENS_DATA_17 0x5A #define EXT_SENS_DATA_18 0x5B #define EXT_SENS_DATA_19 0x5C #define EXT_SENS_DATA_20 0x5D #define EXT_SENS_DATA_21 0x5E #define EXT_SENS_DATA_22 0x5F #define EXT_SENS_DATA_23 0x60 #define MOT_DETECT_STATUS 0x61 #define I2C_SLV0_DO 0x63 #define I2C_SLV1_DO 0x64 #define I2C_SLV2_DO 0x65 #define I2C_SLV3_DO 0x66 #define I2C_MST_DELAY_CTRL 0x67 #define SIGNAL_PATH_RESET 0x68 #define MOT_DETECT_CTRL 0x69 #define USER_CTRL 0x6A // Bit 7 enable DMP, bit 3 reset DMP #define PWR_MGMT_1 0x6B // Device defaults to the SLEEP mode #define PWR_MGMT_2 0x6C #define DMP_BANK 0x6D // Activates a specific bank in the DMP #define DMP_RW_PNT 0x6E // Set read/write pointer to a specific start address in specified DMP bank #define DMP_REG 0x6F // Register in DMP from which to read or to which to write #define DMP_REG_1 0x70 #define DMP_REG_2 0x71 #define FIFO_COUNTH 0x72 #define FIFO_COUNTL 0x73 #define FIFO_R_W 0x74 #define WHO_AM_I_MPU9250 0x75 // Should return 0x71 #define XA_OFFSET_H 0x77 #define XA_OFFSET_L 0x78 #define YA_OFFSET_H 0x7A #define YA_OFFSET_L 0x7B #define ZA_OFFSET_H 0x7D #define ZA_OFFSET_L 0x7E // Using the MSENSR-9250 breakout board, ADO is set to 0 // Seven-bit device address is 110100 for ADO = 0 and 110101 for ADO = 1 //mbed uses the eight-bit device address, so shift seven-bit addresses left by one! #define ADO 0 #if ADO #define MPU9250_ADDRESS 0x69<<1 // Device address when ADO = 1 #else #define MPU9250_ADDRESS 0x68<<1 // Device address when ADO = 0 #endif // Set initial input parameters enum Ascale { AFS_2G = 0, AFS_4G, AFS_8G, AFS_16G }; enum Gscale { GFS_250DPS = 0, GFS_500DPS, GFS_1000DPS, GFS_2000DPS }; enum Mscale { MFS_14BITS = 0, // 0.6 mG per LSB MFS_16BITS // 0.15 mG per LSB }; uint8_t Ascale = AFS_2G; // AFS_2G, AFS_4G, AFS_8G, AFS_16G uint8_t Gscale = GFS_250DPS; // GFS_250DPS, GFS_500DPS, GFS_1000DPS, GFS_2000DPS uint8_t Mscale = MFS_16BITS; // MFS_14BITS or MFS_16BITS, 14-bit or 16-bit magnetometer resolution uint8_t Mmode = 0x06; // Either 8 Hz 0x02) or 100 Hz (0x06) magnetometer data ODR float aRes, gRes, mRes; // scale resolutions per LSB for the sensors int I2Cstate=1; // If I2Cstate!=0, I2C read or write operation has failed //Set up I2C, (SDA,SCL) I2C i2c(I2C_SDA, I2C_SCL); DigitalOut myled(LED1); // Pin definitions int intPin = 12; // These can be changed, 2 and 3 are the Arduinos ext int pins int16_t accelCount[3]; // Stores the 16-bit signed accelerometer sensor output int16_t gyroCount[3]; // Stores the 16-bit signed gyro sensor output int16_t magCount[3]; // Stores the 16-bit signed magnetometer sensor output float magCalibration[3] = {0, 0, 0}, magbias[3] = {0, 0, 0}; // Factory mag calibration and mag bias float gyroBias[3] = {0, 0, 0}, accelBias[3] = {0, 0, 0}; // Bias corrections for gyro and accelerometer float ax, ay, az, gx, gy, gz, mx, my, mz; // variables to hold latest sensor data values int16_t tempCount; // Stores the real internal chip temperature in degrees Celsius float temperature; float SelfTest[6]; float orientation[1]; float magn_x, magn_y; int delt_t = 0; // used to control display output rate int count = 0; // used to control display output rate // parameters for 6 DoF sensor fusion calculations float PI = 3.14159265358979323846f; float GyroMeasError = PI * (60.0f / 180.0f); // gyroscope measurement error in rads/s (start at 60 deg/s), then reduce after ~10 s to 3 float beta = sqrt(3.0f / 4.0f) * GyroMeasError; // compute beta float GyroMeasDrift = PI * (1.0f / 180.0f); // gyroscope measurement drift in rad/s/s (start at 0.0 deg/s/s) float zeta = sqrt(3.0f / 4.0f) * GyroMeasDrift; // compute zeta, the other free parameter in the Madgwick scheme usually set to a small or zero value #define Kp 2.0f * 5.0f // these are the free parameters in the Mahony filter and fusion scheme, Kp for proportional feedback, Ki for integral #define Ki 0.0f float pitch, yaw, roll; float deltat = 0.0f; // integration interval for both filter schemes int lastUpdate = 0, firstUpdate = 0, Now = 0; // used to calculate integration interval // used to calculate integration interval float q[4] = {1.0f, 0.0f, 0.0f, 0.0f}; // vector to hold quaternion float eInt[3] = {0.0f, 0.0f, 0.0f}; // vector to hold integral error for Mahony method class MPU9250 { protected: public: //=================================================================================================================== //====== Set of useful function to access acceleration, gyroscope, and temperature data //=================================================================================================================== void writeByte(uint8_t address, uint8_t subAddress, uint8_t data) { char data_write[2]; data_write[0] = subAddress; data_write[1] = data; I2Cstate = i2c.write(address, data_write, 2, 0); } char readByte(uint8_t address, uint8_t subAddress) { char data[1]; // `data` will store the register data char data_write[1]; data_write[0] = subAddress; I2Cstate = i2c.write(address, data_write, 1, 1); // no stop I2Cstate = i2c.read(address, data, 1, 0); return data[0]; } void readBytes(uint8_t address, uint8_t subAddress, uint8_t count, uint8_t * dest) // count=nº of bytes to read / dest=destiny where data is stored { char data[14]; char data_write[1]; data_write[0] = subAddress; I2Cstate = i2c.write(address, data_write, 1, 1); // no stop I2Cstate = i2c.read(address, data, count, 0); for(int ii = 0; ii < count; ii++) { dest[ii] = data[ii]; } } void getMres() { switch (Mscale) { // Possible magnetometer scales (and their register bit settings) are: // 14 bit resolution (0) and 16 bit resolution (1) case MFS_14BITS: mRes = 10.0*4219.0/8190.0; // Proper scale to return milliGauss break; case MFS_16BITS: mRes = 10.0*4219.0/32760.0; // Proper scale to return milliGauss break; } } void getGres() { switch (Gscale) { // Possible gyro scales (and their register bit settings) are: // 250 DPS (00), 500 DPS (01), 1000 DPS (10), and 2000 DPS (11). // Here's a bit of an algorith to calculate DPS/(ADC tick) based on that 2-bit value: case GFS_250DPS: gRes = 250.0/32768.0; break; case GFS_500DPS: gRes = 500.0/32768.0; break; case GFS_1000DPS: gRes = 1000.0/32768.0; break; case GFS_2000DPS: gRes = 2000.0/32768.0; break; } } void getAres() { switch (Ascale) { // Possible accelerometer scales (and their register bit settings) are: // 2 Gs (00), 4 Gs (01), 8 Gs (10), and 16 Gs (11). // Here's a bit of an algorith to calculate DPS/(ADC tick) based on that 2-bit value: case AFS_2G: aRes = 2.0/32768.0; break; case AFS_4G: aRes = 4.0/32768.0; break; case AFS_8G: aRes = 8.0/32768.0; break; case AFS_16G: aRes = 16.0/32768.0; break; } } void readAccelData(int16_t * destination){ uint8_t rawData[6]; // x/y/z accel register data stored here readBytes(MPU9250_ADDRESS, ACCEL_XOUT_H, 6, &rawData[0]); // Read the six raw data registers into data array destination[0] = (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value destination[1] = (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; destination[2] = (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; } void readGyroData(int16_t * destination){ uint8_t rawData[6]; // x/y/z gyro register data stored here readBytes(MPU9250_ADDRESS, GYRO_XOUT_H, 6, &rawData[0]); // Read the six raw data registers sequentially into data array destination[0] = (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value destination[1] = (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; destination[2] = (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; } void readMagData(int16_t * destination){ uint8_t rawData[7]; // x/y/z gyro register data, ST2 register stored here, must read ST2 at end of data acquisition if(readByte(AK8963_ADDRESS, AK8963_ST1) & 0x01) { // wait for magnetometer data ready bit to be set readBytes(AK8963_ADDRESS, AK8963_XOUT_L, 7, &rawData[0]); // Read the six raw data and ST2 registers sequentially into data array uint8_t c = rawData[6]; // End data read by reading ST2 register if(!(c & 0x08)) { // Check if magnetic sensor overflow set, if not then report data destination[0] = (int16_t)(((int16_t)rawData[1] << 8) | rawData[0]); // Turn the MSB and LSB into a signed 16-bit value destination[1] = (int16_t)(((int16_t)rawData[3] << 8) | rawData[2]) ; // Data stored as little Endian destination[2] = (int16_t)(((int16_t)rawData[5] << 8) | rawData[4]) ; } } } int16_t readTempData(){ uint8_t rawData[2]; // x/y/z gyro register data stored here readBytes(MPU9250_ADDRESS, TEMP_OUT_H, 2, &rawData[0]); // Read the two raw data registers sequentially into data array return (int16_t)(((int16_t)rawData[0]) << 8 | rawData[1]) ; // Turn the MSB and LSB into a 16-bit value } void resetMPU9250(){ // reset device writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x80); // Write a one to bit 7 reset bit; toggle reset device wait(0.1); } void initAK8963(float * destination){ // First extract the factory calibration for each magnetometer axis uint8_t rawData[3]; // x/y/z gyro calibration data stored here writeByte(AK8963_ADDRESS, AK8963_CNTL, 0x00); // Power down magnetometer wait(0.01); writeByte(AK8963_ADDRESS, AK8963_CNTL, 0x0F); // Enter Fuse ROM access mode wait(0.01); readBytes(AK8963_ADDRESS, AK8963_ASAX, 3, &rawData[0]); // Read the x-, y-, and z-axis calibration values destination[0] = (float)(rawData[0] - 128)/256.0f + 1.0f; // Return x-axis sensitivity adjustment values, etc. destination[1] = (float)(rawData[1] - 128)/256.0f + 1.0f; destination[2] = (float)(rawData[2] - 128)/256.0f + 1.0f; writeByte(AK8963_ADDRESS, AK8963_CNTL, 0x00); // Power down magnetometer wait(0.01); // Configure the magnetometer for continuous read and highest resolution // set Mscale bit 4 to 1 (0) to enable 16 (14) bit resolution in CNTL register, // and enable continuous mode data acquisition Mmode (bits [3:0]), 0010 for 8 Hz and 0110 for 100 Hz sample rates writeByte(AK8963_ADDRESS, AK8963_CNTL, Mscale << 4 | Mmode); // Set magnetometer data resolution and sample ODR wait(0.01); } void initMPU9250(){ // Initialize MPU9250 device // wake up device writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x00); // Clear sleep mode bit (6), enable all sensors wait(0.1); // Delay 100 ms for PLL to get established on x-axis gyro; should check for PLL ready interrupt // get stable time source writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x01); // Set clock source to be PLL with x-axis gyroscope reference, bits 2:0 = 001 // Configure Gyro and Accelerometer // Disable FSYNC and set accelerometer and gyro bandwidth to 44 and 42 Hz, respectively; // DLPF_CFG = bits 2:0 = 010; this sets the sample rate at 1 kHz for both // Maximum delay is 4.9 ms which is just over a 200 Hz maximum rate writeByte(MPU9250_ADDRESS, CONFIG, 0x03); // Set sample rate = gyroscope output rate/(1 + SMPLRT_DIV) writeByte(MPU9250_ADDRESS, SMPLRT_DIV, 0x04); // Use a 200 Hz rate; the same rate set in CONFIG above // Set gyroscope full scale range // Range selects FS_SEL and AFS_SEL are 0 - 3, so 2-bit values are left-shifted into positions 4:3 uint8_t c = readByte(MPU9250_ADDRESS, GYRO_CONFIG); writeByte(MPU9250_ADDRESS, GYRO_CONFIG, c & ~0xE0); // Clear self-test bits [7:5] writeByte(MPU9250_ADDRESS, GYRO_CONFIG, c & ~0x18); // Clear AFS bits [4:3] writeByte(MPU9250_ADDRESS, GYRO_CONFIG, c | Gscale << 3); // Set full scale range for the gyro // Set accelerometer configuration c = readByte(MPU9250_ADDRESS, ACCEL_CONFIG); writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, c & ~0xE0); // Clear self-test bits [7:5] writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, c & ~0x18); // Clear AFS bits [4:3] writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, c | Ascale << 3); // Set full scale range for the accelerometer // Set accelerometer sample rate configuration // It is possible to get a 4 kHz sample rate from the accelerometer by choosing 1 for // accel_fchoice_b bit [3]; in this case the bandwidth is 1.13 kHz c = readByte(MPU9250_ADDRESS, ACCEL_CONFIG2); writeByte(MPU9250_ADDRESS, ACCEL_CONFIG2, c & ~0x0F); // Clear accel_fchoice_b (bit 3) and A_DLPFG (bits [2:0]) writeByte(MPU9250_ADDRESS, ACCEL_CONFIG2, c | 0x03); // Set accelerometer rate to 1 kHz and bandwidth to 41 Hz // The accelerometer, gyro, and thermometer are set to 1 kHz sample rates, // but all these rates are further reduced by a factor of 5 to 200 Hz because of the SMPLRT_DIV setting // Configure Interrupts and Bypass Enable // Set interrupt pin active high, push-pull, and clear on read of INT_STATUS, enable I2C_BYPASS_EN so additional chips // can join the I2C bus and all can be controlled by the Arduino as master writeByte(MPU9250_ADDRESS, INT_PIN_CFG, 0x22); writeByte(MPU9250_ADDRESS, INT_ENABLE, 0x01); // Enable data ready (bit 0) interrupt } // Function which accumulates gyro and accelerometer data after device initialization. It calculates the average // of the at-rest readings and then loads the resulting offsets into accelerometer and gyro bias registers. void calibrateMPU9250(float * dest1, float * dest2) { uint8_t data[12]; // data array to hold accelerometer and gyro x, y, z, data uint16_t ii, packet_count, fifo_count; int32_t gyro_bias[3] = {0, 0, 0}, accel_bias[3] = {0, 0, 0}; // reset device, reset all registers, clear gyro and accelerometer bias registers writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x80); // Write a one to bit 7 reset bit; toggle reset device wait(0.1); // get stable time source // Set clock source to be PLL with x-axis gyroscope reference, bits 2:0 = 001 writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x01); writeByte(MPU9250_ADDRESS, PWR_MGMT_2, 0x00); wait(0.2); // Configure device for bias calculation writeByte(MPU9250_ADDRESS, INT_ENABLE, 0x00); // Disable all interrupts writeByte(MPU9250_ADDRESS, FIFO_EN, 0x00); // Disable FIFO writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x00); // Turn on internal clock source writeByte(MPU9250_ADDRESS, I2C_MST_CTRL, 0x00); // Disable I2C master writeByte(MPU9250_ADDRESS, USER_CTRL, 0x00); // Disable FIFO and I2C master modes writeByte(MPU9250_ADDRESS, USER_CTRL, 0x0C); // Reset FIFO and DMP wait(0.015); // Configure MPU9250 gyro and accelerometer for bias calculation writeByte(MPU9250_ADDRESS, CONFIG, 0x01); // Set low-pass filter to 188 Hz writeByte(MPU9250_ADDRESS, SMPLRT_DIV, 0x00); // Set sample rate to 1 kHz writeByte(MPU9250_ADDRESS, GYRO_CONFIG, 0x00); // Set gyro full-scale to 250 degrees per second, maximum sensitivity writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, 0x00); // Set accelerometer full-scale to 2 g, maximum sensitivity uint16_t gyrosensitivity = 131; // = 131 LSB/degrees/sec uint16_t accelsensitivity = 16384; // = 16384 LSB/g // Configure FIFO to capture accelerometer and gyro data. This data will be used for bias calculation writeByte(MPU9250_ADDRESS, USER_CTRL, 0x40); // Enable FIFO writeByte(MPU9250_ADDRESS, FIFO_EN, 0x78); // Enable gyro and accelerometer sensors for FIFO (max size 512 bytes in MPU-9250) wait(0.04); // accumulate 40 samples in 80 milliseconds = 480 bytes // At end of sample accumulation, turn off FIFO sensor read writeByte(MPU9250_ADDRESS, FIFO_EN, 0x00); // Disable gyro and accelerometer sensors for FIFO readBytes(MPU9250_ADDRESS, FIFO_COUNTH, 2, &data[0]); // read FIFO sample count fifo_count = ((uint16_t)data[0] << 8) | data[1]; packet_count = fifo_count/12;// How many sets of full gyro and accelerometer data for averaging for (ii = 0; ii < packet_count; ii++) { int16_t accel_temp[3] = {0, 0, 0}, gyro_temp[3] = {0, 0, 0}; readBytes(MPU9250_ADDRESS, FIFO_R_W, 12, &data[0]); // read data for averaging accel_temp[0] = (int16_t) (((int16_t)data[0] << 8) | data[1] ) ; // Form signed 16-bit integer for each sample in FIFO accel_temp[1] = (int16_t) (((int16_t)data[2] << 8) | data[3] ) ; accel_temp[2] = (int16_t) (((int16_t)data[4] << 8) | data[5] ) ; gyro_temp[0] = (int16_t) (((int16_t)data[6] << 8) | data[7] ) ; gyro_temp[1] = (int16_t) (((int16_t)data[8] << 8) | data[9] ) ; gyro_temp[2] = (int16_t) (((int16_t)data[10] << 8) | data[11]) ; accel_bias[0] += (int32_t) accel_temp[0]; // Sum individual signed 16-bit biases to get accumulated signed 32-bit biases accel_bias[1] += (int32_t) accel_temp[1]; accel_bias[2] += (int32_t) accel_temp[2]; gyro_bias[0] += (int32_t) gyro_temp[0]; gyro_bias[1] += (int32_t) gyro_temp[1]; gyro_bias[2] += (int32_t) gyro_temp[2]; } accel_bias[0] /= (int32_t) packet_count; // Normalize sums to get average count biases accel_bias[1] /= (int32_t) packet_count; accel_bias[2] /= (int32_t) packet_count; gyro_bias[0] /= (int32_t) packet_count; gyro_bias[1] /= (int32_t) packet_count; gyro_bias[2] /= (int32_t) packet_count; if(accel_bias[2] > 0L) {accel_bias[2] -= (int32_t) accelsensitivity;} // Remove gravity from the z-axis accelerometer bias calculation else {accel_bias[2] += (int32_t) accelsensitivity;} // Construct the gyro biases for push to the hardware gyro bias registers, which are reset to zero upon device startup data[0] = (-gyro_bias[0]/4 >> 8) & 0xFF; // Divide by 4 to get 32.9 LSB per deg/s to conform to expected bias input format data[1] = (-gyro_bias[0]/4) & 0xFF; // Biases are additive, so change sign on calculated average gyro biases data[2] = (-gyro_bias[1]/4 >> 8) & 0xFF; data[3] = (-gyro_bias[1]/4) & 0xFF; data[4] = (-gyro_bias[2]/4 >> 8) & 0xFF; data[5] = (-gyro_bias[2]/4) & 0xFF; /// Push gyro biases to hardware registers /* writeByte(MPU9250_ADDRESS, XG_OFFSET_H, data[0]); writeByte(MPU9250_ADDRESS, XG_OFFSET_L, data[1]); writeByte(MPU9250_ADDRESS, YG_OFFSET_H, data[2]); writeByte(MPU9250_ADDRESS, YG_OFFSET_L, data[3]); writeByte(MPU9250_ADDRESS, ZG_OFFSET_H, data[4]); writeByte(MPU9250_ADDRESS, ZG_OFFSET_L, data[5]); */ dest1[0] = (float) gyro_bias[0]/(float) gyrosensitivity; // construct gyro bias in deg/s for later manual subtraction dest1[1] = (float) gyro_bias[1]/(float) gyrosensitivity; dest1[2] = (float) gyro_bias[2]/(float) gyrosensitivity; // Construct the accelerometer biases for push to the hardware accelerometer bias registers. These registers contain // factory trim values which must be added to the calculated accelerometer biases; on boot up these registers will hold // non-zero values. In addition, bit 0 of the lower byte must be preserved since it is used for temperature // compensation calculations. Accelerometer bias registers expect bias input as 2048 LSB per g, so that // the accelerometer biases calculated above must be divided by 8. int32_t accel_bias_reg[3] = {0, 0, 0}; // A place to hold the factory accelerometer trim biases readBytes(MPU9250_ADDRESS, XA_OFFSET_H, 2, &data[0]); // Read factory accelerometer trim values accel_bias_reg[0] = (int16_t) ((int16_t)data[0] << 8) | data[1]; readBytes(MPU9250_ADDRESS, YA_OFFSET_H, 2, &data[0]); accel_bias_reg[1] = (int16_t) ((int16_t)data[0] << 8) | data[1]; readBytes(MPU9250_ADDRESS, ZA_OFFSET_H, 2, &data[0]); accel_bias_reg[2] = (int16_t) ((int16_t)data[0] << 8) | data[1]; uint32_t mask = 1uL; // Define mask for temperature compensation bit 0 of lower byte of accelerometer bias registers uint8_t mask_bit[3] = {0, 0, 0}; // Define array to hold mask bit for each accelerometer bias axis for(ii = 0; ii < 3; ii++) { if(accel_bias_reg[ii] & mask) mask_bit[ii] = 0x01; // If temperature compensation bit is set, record that fact in mask_bit } // Construct total accelerometer bias, including calculated average accelerometer bias from above accel_bias_reg[0] -= (accel_bias[0]/8); // Subtract calculated averaged accelerometer bias scaled to 2048 LSB/g (16 g full scale) accel_bias_reg[1] -= (accel_bias[1]/8); accel_bias_reg[2] -= (accel_bias[2]/8); data[0] = (accel_bias_reg[0] >> 8) & 0xFF; data[1] = (accel_bias_reg[0]) & 0xFF; data[1] = data[1] | mask_bit[0]; // preserve temperature compensation bit when writing back to accelerometer bias registers data[2] = (accel_bias_reg[1] >> 8) & 0xFF; data[3] = (accel_bias_reg[1]) & 0xFF; data[3] = data[3] | mask_bit[1]; // preserve temperature compensation bit when writing back to accelerometer bias registers data[4] = (accel_bias_reg[2] >> 8) & 0xFF; data[5] = (accel_bias_reg[2]) & 0xFF; data[5] = data[5] | mask_bit[2]; // preserve temperature compensation bit when writing back to accelerometer bias registers // Apparently this is not working for the acceleration biases in the MPU-9250 // Are we handling the temperature correction bit properly? // Push accelerometer biases to hardware registers /* writeByte(MPU9250_ADDRESS, XA_OFFSET_H, data[0]); writeByte(MPU9250_ADDRESS, XA_OFFSET_L, data[1]); writeByte(MPU9250_ADDRESS, YA_OFFSET_H, data[2]); writeByte(MPU9250_ADDRESS, YA_OFFSET_L, data[3]); writeByte(MPU9250_ADDRESS, ZA_OFFSET_H, data[4]); writeByte(MPU9250_ADDRESS, ZA_OFFSET_L, data[5]); */ // Output scaled accelerometer biases for manual subtraction in the main program dest2[0] = (float)accel_bias[0]/(float)accelsensitivity; dest2[1] = (float)accel_bias[1]/(float)accelsensitivity; dest2[2] = (float)accel_bias[2]/(float)accelsensitivity; } // Accelerometer and gyroscope self test; check calibration wrt factory settings void MPU9250SelfTest(float * destination) // Should return percent deviation from factory trim values, +/- 14 or less deviation is a pass { uint8_t rawData[6] = {0, 0, 0, 0, 0, 0}; uint8_t selfTest[6]; int16_t gAvg[3], aAvg[3], aSTAvg[3], gSTAvg[3]; float factoryTrim[6]; uint8_t FS = 0; writeByte(MPU9250_ADDRESS, SMPLRT_DIV, 0x00); // Set gyro sample rate to 1 kHz writeByte(MPU9250_ADDRESS, CONFIG, 0x02); // Set gyro sample rate to 1 kHz and DLPF to 92 Hz writeByte(MPU9250_ADDRESS, GYRO_CONFIG, 1<<FS); // Set full scale range for the gyro to 250 dps writeByte(MPU9250_ADDRESS, ACCEL_CONFIG2, 0x02); // Set accelerometer rate to 1 kHz and bandwidth to 92 Hz writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, 1<<FS); // Set full scale range for the accelerometer to 2 g for( int ii = 0; ii < 200; ii++) { // get average current values of gyro and acclerometer readBytes(MPU9250_ADDRESS, ACCEL_XOUT_H, 6, &rawData[0]); // Read the six raw data registers into data array aAvg[0] += (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value aAvg[1] += (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; aAvg[2] += (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; readBytes(MPU9250_ADDRESS, GYRO_XOUT_H, 6, &rawData[0]); // Read the six raw data registers sequentially into data array gAvg[0] += (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value gAvg[1] += (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; gAvg[2] += (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; } for (int ii =0; ii < 3; ii++) { // Get average of 200 values and store as average current readings aAvg[ii] /= 200; gAvg[ii] /= 200; } // Configure the accelerometer for self-test writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, 0xE0); // Enable self test on all three axes and set accelerometer range to +/- 2 g writeByte(MPU9250_ADDRESS, GYRO_CONFIG, 0xE0); // Enable self test on all three axes and set gyro range to +/- 250 degrees/s wait_ms(25); // Delay a while to let the device stabilize for( int ii = 0; ii < 200; ii++) { // get average self-test values of gyro and acclerometer readBytes(MPU9250_ADDRESS, ACCEL_XOUT_H, 6, &rawData[0]); // Read the six raw data registers into data array aSTAvg[0] += (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value aSTAvg[1] += (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; aSTAvg[2] += (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; readBytes(MPU9250_ADDRESS, GYRO_XOUT_H, 6, &rawData[0]); // Read the six raw data registers sequentially into data array gSTAvg[0] += (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value gSTAvg[1] += (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; gSTAvg[2] += (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; } for (int ii =0; ii < 3; ii++) { // Get average of 200 values and store as average self-test readings aSTAvg[ii] /= 200; gSTAvg[ii] /= 200; } // Configure the gyro and accelerometer for normal operation writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, 0x00); writeByte(MPU9250_ADDRESS, GYRO_CONFIG, 0x00); //delay(25); // Delay a while to let the device stabilize wait_ms(25); // Delay a while to let the device stabilize // Retrieve accelerometer and gyro factory Self-Test Code from USR_Reg selfTest[0] = readByte(MPU9250_ADDRESS, SELF_TEST_X_ACCEL); // X-axis accel self-test results selfTest[1] = readByte(MPU9250_ADDRESS, SELF_TEST_Y_ACCEL); // Y-axis accel self-test results selfTest[2] = readByte(MPU9250_ADDRESS, SELF_TEST_Z_ACCEL); // Z-axis accel self-test results selfTest[3] = readByte(MPU9250_ADDRESS, SELF_TEST_X_GYRO); // X-axis gyro self-test results selfTest[4] = readByte(MPU9250_ADDRESS, SELF_TEST_Y_GYRO); // Y-axis gyro self-test results selfTest[5] = readByte(MPU9250_ADDRESS, SELF_TEST_Z_GYRO); // Z-axis gyro self-test results // Retrieve factory self-test value from self-test code reads factoryTrim[0] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[0] - 1.0) )); // FT[Xa] factory trim calculation factoryTrim[1] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[1] - 1.0) )); // FT[Ya] factory trim calculation factoryTrim[2] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[2] - 1.0) )); // FT[Za] factory trim calculation factoryTrim[3] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[3] - 1.0) )); // FT[Xg] factory trim calculation factoryTrim[4] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[4] - 1.0) )); // FT[Yg] factory trim calculation factoryTrim[5] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[5] - 1.0) )); // FT[Zg] factory trim calculation // Report results as a ratio of (STR - FT)/FT; the change from Factory Trim of the Self-Test Response // To get percent, must multiply by 100 for (int i = 0; i < 3; i++) { destination[i] = 100.0*((float)(aSTAvg[i] - aAvg[i]))/factoryTrim[i]; // Report percent differences destination[i+3] = 100.0*((float)(gSTAvg[i] - gAvg[i]))/factoryTrim[i+3]; // Report percent differences } } void getCompassOrientation(float * orient){ // Obtains the orientation of the device in degrees. 0 degrees North. 180 degrees South. /* Remember that it is the earth's rotational axis that defines the geographic north and south poles that we use for map references. It turns out that there is a discrepancy of about 11.5 degrees between the geographic poles and the magnetic poles. The last is what the magnetometer will read. A value, called the declination angle, can be applied to the magnetic direction to correct for this. On Valencia (Spain) this value is about 0 degrees. */ // First of all measure 3 axis magnetometer values (only X and Y axis is used): readMagData(magCount); // Read the x/y/z adc values // Calculate the magnetometer values in milliGauss // Include factory calibration per data sheet and user environmental corrections if (I2Cstate == 0){ // no error on I2C I2Cstate = 1; magn_x = (float)magCount[0]*mRes*magCalibration[0] - magbias[0]; // get actual magnetometer value, this depends on scale being set magn_y = (float)magCount[1]*mRes*magCalibration[1] - magbias[1]; } // Now obtains the orientation value: if (magn_y>0) orient[0] = 90.0 - (float) ( atan(magn_x/magn_y)*180/M_PI ); else if (magn_y<0) orient[0] = 270.0 - (float) ( atan(magn_x/magn_y)*180/M_PI ); else if (magn_y == 0){ if (magn_x<0) orient[0] = 180.0; else orient[0] = 0.0; } } // Implementation of Sebastian Madgwick's "...efficient orientation filter for... inertial/magnetic sensor arrays" // (see http://www.x-io.co.uk/category/open-source/ for examples and more details) // which fuses acceleration, rotation rate, and magnetic moments to produce a quaternion-based estimate of absolute // device orientation -- which can be converted to yaw, pitch, and roll. Useful for stabilizing quadcopters, etc. // The performance of the orientation filter is at least as good as conventional Kalman-based filtering algorithms // but is much less computationally intensive---it can be performed on a 3.3 V Pro Mini operating at 8 MHz! void MadgwickQuaternionUpdate(float ax, float ay, float az, float gx, float gy, float gz, float mx, float my, float mz) { float q1 = q[0], q2 = q[1], q3 = q[2], q4 = q[3]; // short name local variable for readability float norm; float hx, hy, _2bx, _2bz; float s1, s2, s3, s4; float qDot1, qDot2, qDot3, qDot4; // Auxiliary variables to avoid repeated arithmetic float _2q1mx; float _2q1my; float _2q1mz; float _2q2mx; float _4bx; float _4bz; float _2q1 = 2.0f * q1; float _2q2 = 2.0f * q2; float _2q3 = 2.0f * q3; float _2q4 = 2.0f * q4; float _2q1q3 = 2.0f * q1 * q3; float _2q3q4 = 2.0f * q3 * q4; float q1q1 = q1 * q1; float q1q2 = q1 * q2; float q1q3 = q1 * q3; float q1q4 = q1 * q4; float q2q2 = q2 * q2; float q2q3 = q2 * q3; float q2q4 = q2 * q4; float q3q3 = q3 * q3; float q3q4 = q3 * q4; float q4q4 = q4 * q4; // Normalise accelerometer measurement norm = sqrt(ax * ax + ay * ay + az * az); if (norm == 0.0f) return; // handle NaN norm = 1.0f/norm; ax *= norm; ay *= norm; az *= norm; // Normalise magnetometer measurement norm = sqrt(mx * mx + my * my + mz * mz); if (norm == 0.0f) return; // handle NaN norm = 1.0f/norm; mx *= norm; my *= norm; mz *= norm; // Reference direction of Earth's magnetic field _2q1mx = 2.0f * q1 * mx; _2q1my = 2.0f * q1 * my; _2q1mz = 2.0f * q1 * mz; _2q2mx = 2.0f * q2 * mx; hx = mx * q1q1 - _2q1my * q4 + _2q1mz * q3 + mx * q2q2 + _2q2 * my * q3 + _2q2 * mz * q4 - mx * q3q3 - mx * q4q4; hy = _2q1mx * q4 + my * q1q1 - _2q1mz * q2 + _2q2mx * q3 - my * q2q2 + my * q3q3 + _2q3 * mz * q4 - my * q4q4; _2bx = sqrt(hx * hx + hy * hy); _2bz = -_2q1mx * q3 + _2q1my * q2 + mz * q1q1 + _2q2mx * q4 - mz * q2q2 + _2q3 * my * q4 - mz * q3q3 + mz * q4q4; _4bx = 2.0f * _2bx; _4bz = 2.0f * _2bz; // Gradient decent algorithm corrective step s1 = -_2q3 * (2.0f * q2q4 - _2q1q3 - ax) + _2q2 * (2.0f * q1q2 + _2q3q4 - ay) - _2bz * q3 * (_2bx * (0.5f - q3q3 - q4q4) + _2bz * (q2q4 - q1q3) - mx) + (-_2bx * q4 + _2bz * q2) * (_2bx * (q2q3 - q1q4) + _2bz * (q1q2 + q3q4) - my) + _2bx * q3 * (_2bx * (q1q3 + q2q4) + _2bz * (0.5f - q2q2 - q3q3) - mz); s2 = _2q4 * (2.0f * q2q4 - _2q1q3 - ax) + _2q1 * (2.0f * q1q2 + _2q3q4 - ay) - 4.0f * q2 * (1.0f - 2.0f * q2q2 - 2.0f * q3q3 - az) + _2bz * q4 * (_2bx * (0.5f - q3q3 - q4q4) + _2bz * (q2q4 - q1q3) - mx) + (_2bx * q3 + _2bz * q1) * (_2bx * (q2q3 - q1q4) + _2bz * (q1q2 + q3q4) - my) + (_2bx * q4 - _4bz * q2) * (_2bx * (q1q3 + q2q4) + _2bz * (0.5f - q2q2 - q3q3) - mz); s3 = -_2q1 * (2.0f * q2q4 - _2q1q3 - ax) + _2q4 * (2.0f * q1q2 + _2q3q4 - ay) - 4.0f * q3 * (1.0f - 2.0f * q2q2 - 2.0f * q3q3 - az) + (-_4bx * q3 - _2bz * q1) * (_2bx * (0.5f - q3q3 - q4q4) + _2bz * (q2q4 - q1q3) - mx) + (_2bx * q2 + _2bz * q4) * (_2bx * (q2q3 - q1q4) + _2bz * (q1q2 + q3q4) - my) + (_2bx * q1 - _4bz * q3) * (_2bx * (q1q3 + q2q4) + _2bz * (0.5f - q2q2 - q3q3) - mz); s4 = _2q2 * (2.0f * q2q4 - _2q1q3 - ax) + _2q3 * (2.0f * q1q2 + _2q3q4 - ay) + (-_4bx * q4 + _2bz * q2) * (_2bx * (0.5f - q3q3 - q4q4) + _2bz * (q2q4 - q1q3) - mx) + (-_2bx * q1 + _2bz * q3) * (_2bx * (q2q3 - q1q4) + _2bz * (q1q2 + q3q4) - my) + _2bx * q2 * (_2bx * (q1q3 + q2q4) + _2bz * (0.5f - q2q2 - q3q3) - mz); norm = sqrt(s1 * s1 + s2 * s2 + s3 * s3 + s4 * s4); // normalise step magnitude norm = 1.0f/norm; s1 *= norm; s2 *= norm; s3 *= norm; s4 *= norm; // Compute rate of change of quaternion qDot1 = 0.5f * (-q2 * gx - q3 * gy - q4 * gz) - beta * s1; qDot2 = 0.5f * (q1 * gx + q3 * gz - q4 * gy) - beta * s2; qDot3 = 0.5f * (q1 * gy - q2 * gz + q4 * gx) - beta * s3; qDot4 = 0.5f * (q1 * gz + q2 * gy - q3 * gx) - beta * s4; // Integrate to yield quaternion q1 += qDot1 * deltat; q2 += qDot2 * deltat; q3 += qDot3 * deltat; q4 += qDot4 * deltat; norm = sqrt(q1 * q1 + q2 * q2 + q3 * q3 + q4 * q4); // normalise quaternion norm = 1.0f/norm; q[0] = q1 * norm; q[1] = q2 * norm; q[2] = q3 * norm; q[3] = q4 * norm; } // Similar to Madgwick scheme but uses proportional and integral filtering on the error between estimated reference vectors and // measured ones. void MahonyQuaternionUpdate(float ax, float ay, float az, float gx, float gy, float gz, float mx, float my, float mz) { float q1 = q[0], q2 = q[1], q3 = q[2], q4 = q[3]; // short name local variable for readability float norm; float hx, hy, bx, bz; float vx, vy, vz, wx, wy, wz; float ex, ey, ez; float pa, pb, pc; // Auxiliary variables to avoid repeated arithmetic float q1q1 = q1 * q1; float q1q2 = q1 * q2; float q1q3 = q1 * q3; float q1q4 = q1 * q4; float q2q2 = q2 * q2; float q2q3 = q2 * q3; float q2q4 = q2 * q4; float q3q3 = q3 * q3; float q3q4 = q3 * q4; float q4q4 = q4 * q4; // Normalise accelerometer measurement norm = sqrt(ax * ax + ay * ay + az * az); if (norm == 0.0f) return; // handle NaN norm = 1.0f / norm; // use reciprocal for division ax *= norm; ay *= norm; az *= norm; // Normalise magnetometer measurement norm = sqrt(mx * mx + my * my + mz * mz); if (norm == 0.0f) return; // handle NaN norm = 1.0f / norm; // use reciprocal for division mx *= norm; my *= norm; mz *= norm; // Reference direction of Earth's magnetic field hx = 2.0f * mx * (0.5f - q3q3 - q4q4) + 2.0f * my * (q2q3 - q1q4) + 2.0f * mz * (q2q4 + q1q3); hy = 2.0f * mx * (q2q3 + q1q4) + 2.0f * my * (0.5f - q2q2 - q4q4) + 2.0f * mz * (q3q4 - q1q2); bx = sqrt((hx * hx) + (hy * hy)); bz = 2.0f * mx * (q2q4 - q1q3) + 2.0f * my * (q3q4 + q1q2) + 2.0f * mz * (0.5f - q2q2 - q3q3); // Estimated direction of gravity and magnetic field vx = 2.0f * (q2q4 - q1q3); vy = 2.0f * (q1q2 + q3q4); vz = q1q1 - q2q2 - q3q3 + q4q4; wx = 2.0f * bx * (0.5f - q3q3 - q4q4) + 2.0f * bz * (q2q4 - q1q3); wy = 2.0f * bx * (q2q3 - q1q4) + 2.0f * bz * (q1q2 + q3q4); wz = 2.0f * bx * (q1q3 + q2q4) + 2.0f * bz * (0.5f - q2q2 - q3q3); // Error is cross product between estimated direction and measured direction of gravity ex = (ay * vz - az * vy) + (my * wz - mz * wy); ey = (az * vx - ax * vz) + (mz * wx - mx * wz); ez = (ax * vy - ay * vx) + (mx * wy - my * wx); if (Ki > 0.0f) { eInt[0] += ex; // accumulate integral error eInt[1] += ey; eInt[2] += ez; } else { eInt[0] = 0.0f; // prevent integral wind up eInt[1] = 0.0f; eInt[2] = 0.0f; } // Apply feedback terms gx = gx + Kp * ex + Ki * eInt[0]; gy = gy + Kp * ey + Ki * eInt[1]; gz = gz + Kp * ez + Ki * eInt[2]; // Integrate rate of change of quaternion pa = q2; pb = q3; pc = q4; q1 = q1 + (-q2 * gx - q3 * gy - q4 * gz) * (0.5f * deltat); q2 = pa + (q1 * gx + pb * gz - pc * gy) * (0.5f * deltat); q3 = pb + (q1 * gy - pa * gz + pc * gx) * (0.5f * deltat); q4 = pc + (q1 * gz + pa * gy - pb * gx) * (0.5f * deltat); // Normalise quaternion norm = sqrt(q1 * q1 + q2 * q2 + q3 * q3 + q4 * q4); norm = 1.0f / norm; q[0] = q1 * norm; q[1] = q2 * norm; q[2] = q3 * norm; q[3] = q4 * norm; } }; #endif