Important changes to repositories hosted on mbed.com
Mbed hosted mercurial repositories are deprecated and are due to be permanently deleted in July 2026.
To keep a copy of this software download the repository Zip archive or clone locally using Mercurial.
It is also possible to export all your personal repositories from the account settings page.
Dependencies: mbed GPSINT SDFileSystem
MPU9250.cpp
00001 #include "MPU9250.h" 00002 00003 #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 00004 #define Ki 0.0f 00005 00006 //****************************************************************************** 00007 MPU9250::MPU9250(PinName sda, PinName scl) 00008 { 00009 i2c_ = new I2C(sda, scl); 00010 i2c_->frequency(400000); 00011 } 00012 00013 //****************************************************************************** 00014 MPU9250::MPU9250(I2C *i2c):i2c_(i2c){} 00015 00016 //****************************************************************************** 00017 MPU9250::~MPU9250() 00018 { 00019 delete i2c_; 00020 } 00021 00022 void MPU9250::writeByte(uint8_t address, uint8_t subAddress, uint8_t data) 00023 { 00024 char data_write[2]; 00025 data_write[0] = subAddress; 00026 data_write[1] = data; 00027 i2c_->write(address, data_write, 2, 0); 00028 } 00029 00030 char MPU9250::readByte(uint8_t address, uint8_t subAddress) 00031 { 00032 char data[1]; // `data` will store the register data 00033 char data_write[1]; 00034 data_write[0] = subAddress; 00035 i2c_->write(address, data_write, 1, 1); // no stop 00036 i2c_->read(address, data, 1, 0); 00037 return data[0]; 00038 } 00039 00040 void MPU9250::readBytes(uint8_t address, uint8_t subAddress, uint8_t count, uint8_t * dest) 00041 { 00042 char data[14]; 00043 char data_write[1]; 00044 data_write[0] = subAddress; 00045 i2c_->write(address, data_write, 1, 1); // no stop 00046 i2c_->read(address, data, count, 0); 00047 for(int ii = 0; ii < count; ii++) { 00048 dest[ii] = data[ii]; 00049 } 00050 } 00051 00052 void MPU9250::getMres() { 00053 switch (Mscale) 00054 { 00055 // Possible magnetometer scales (and their register bit settings) are: 00056 // 14 bit resolution (0) and 16 bit resolution (1) 00057 case MFS_14BITS: 00058 mRes = 10.0*4219.0/8190.0; // Proper scale to return milliGauss 00059 break; 00060 case MFS_16BITS: 00061 mRes = 10.0*4219.0/32760.0; // Proper scale to return milliGauss 00062 break; 00063 } 00064 } 00065 00066 00067 void MPU9250::getGres() { 00068 switch (Gscale) 00069 { 00070 // Possible gyro scales (and their register bit settings) are: 00071 // 250 DPS (00), 500 DPS (01), 1000 DPS (10), and 2000 DPS (11). 00072 // Here's a bit of an algorith to calculate DPS/(ADC tick) based on that 2-bit value: 00073 case GFS_250DPS: 00074 gRes = 250.0/32768.0; 00075 break; 00076 case GFS_500DPS: 00077 gRes = 500.0/32768.0; 00078 break; 00079 case GFS_1000DPS: 00080 gRes = 1000.0/32768.0; 00081 break; 00082 case GFS_2000DPS: 00083 gRes = 2000.0/32768.0; 00084 break; 00085 } 00086 } 00087 00088 00089 void MPU9250::getAres() { 00090 switch (Ascale) 00091 { 00092 // Possible accelerometer scales (and their register bit settings) are: 00093 // 2 Gs (00), 4 Gs (01), 8 Gs (10), and 16 Gs (11). 00094 // Here's a bit of an algorith to calculate DPS/(ADC tick) based on that 2-bit value: 00095 case AFS_2G: 00096 aRes = 2.0/32768.0; 00097 break; 00098 case AFS_4G: 00099 aRes = 4.0/32768.0; 00100 break; 00101 case AFS_8G: 00102 aRes = 8.0/32768.0; 00103 break; 00104 case AFS_16G: 00105 aRes = 16.0/32768.0; 00106 break; 00107 } 00108 } 00109 00110 00111 void MPU9250::readAccelData(int16_t * destination) 00112 { 00113 uint8_t rawData[6]; // x/y/z accel register data stored here 00114 readBytes(MPU9250_ADDRESS, ACCEL_XOUT_H, 6, &rawData[0]); // Read the six raw data registers into data array 00115 destination[0] = (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value 00116 destination[1] = (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; 00117 destination[2] = (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; 00118 } 00119 00120 void MPU9250::readGyroData(int16_t * destination) 00121 { 00122 uint8_t rawData[6]; // x/y/z gyro register data stored here 00123 readBytes(MPU9250_ADDRESS, GYRO_XOUT_H, 6, &rawData[0]); // Read the six raw data registers sequentially into data array 00124 destination[0] = (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value 00125 destination[1] = (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; 00126 destination[2] = (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; 00127 } 00128 00129 void MPU9250::readMagData(int16_t * destination) 00130 { 00131 uint8_t rawData[7]; // x/y/z gyro register data, ST2 register stored here, must read ST2 at end of data acquisition 00132 if(readByte(AK8963_ADDRESS, AK8963_ST1) & 0x01) { // wait for magnetometer data ready bit to be set 00133 readBytes(AK8963_ADDRESS, AK8963_XOUT_L, 7, &rawData[0]); // Read the six raw data and ST2 registers sequentially into data array 00134 uint8_t c = rawData[6]; // End data read by reading ST2 register 00135 if(!(c & 0x08)) { // Check if magnetic sensor overflow set, if not then report data 00136 destination[0] = (int16_t)(((int16_t)rawData[1] << 8) | rawData[0]); // Turn the MSB and LSB into a signed 16-bit value 00137 destination[1] = (int16_t)(((int16_t)rawData[3] << 8) | rawData[2]) ; // Data stored as little Endian 00138 destination[2] = (int16_t)(((int16_t)rawData[5] << 8) | rawData[4]) ; 00139 } 00140 } 00141 } 00142 00143 int16_t MPU9250::readTempData() 00144 { 00145 uint8_t rawData[2]; // x/y/z gyro register data stored here 00146 readBytes(MPU9250_ADDRESS, TEMP_OUT_H, 2, &rawData[0]); // Read the two raw data registers sequentially into data array 00147 return (int16_t)(((int16_t)rawData[0]) << 8 | rawData[1]) ; // Turn the MSB and LSB into a 16-bit value 00148 } 00149 00150 00151 void MPU9250::resetMPU9250() { 00152 // reset device 00153 writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x80); // Write a one to bit 7 reset bit; toggle reset device 00154 wait(0.1); 00155 } 00156 00157 void MPU9250::initAK8963(float * destination) 00158 { 00159 // First extract the factory calibration for each magnetometer axis 00160 uint8_t rawData[3]; // x/y/z gyro calibration data stored here 00161 writeByte(AK8963_ADDRESS, AK8963_CNTL, 0x00); // Power down magnetometer 00162 wait(0.01); 00163 writeByte(AK8963_ADDRESS, AK8963_CNTL, 0x0F); // Enter Fuse ROM access mode 00164 wait(0.01); 00165 readBytes(AK8963_ADDRESS, AK8963_ASAX, 3, &rawData[0]); // Read the x-, y-, and z-axis calibration values 00166 destination[0] = (float)(rawData[0] - 128)/256.0f + 1.0f; // Return x-axis sensitivity adjustment values, etc. 00167 destination[1] = (float)(rawData[1] - 128)/256.0f + 1.0f; 00168 destination[2] = (float)(rawData[2] - 128)/256.0f + 1.0f; 00169 writeByte(AK8963_ADDRESS, AK8963_CNTL, 0x00); // Power down magnetometer 00170 wait(0.01); 00171 // Configure the magnetometer for continuous read and highest resolution 00172 // set Mscale bit 4 to 1 (0) to enable 16 (14) bit resolution in CNTL register, 00173 // and enable continuous mode data acquisition Mmode (bits [3:0]), 0010 for 8 Hz and 0110 for 100 Hz sample rates 00174 writeByte(AK8963_ADDRESS, AK8963_CNTL, Mscale << 4 | Mmode); // Set magnetometer data resolution and sample ODR 00175 wait(0.01); 00176 } 00177 00178 00179 void MPU9250::initMPU9250() 00180 { 00181 // Initialize MPU9250 device 00182 // wake up device 00183 writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x00); // Clear sleep mode bit (6), enable all sensors 00184 wait(0.1); // Delay 100 ms for PLL to get established on x-axis gyro; should check for PLL ready interrupt 00185 00186 // get stable time source 00187 writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x01); // Set clock source to be PLL with x-axis gyroscope reference, bits 2:0 = 001 00188 00189 // Configure Gyro and Accelerometer 00190 // Disable FSYNC and set accelerometer and gyro bandwidth to 44 and 42 Hz, respectively; 00191 // DLPF_CFG = bits 2:0 = 010; this sets the sample rate at 1 kHz for both 00192 // Maximum delay is 4.9 ms which is just over a 200 Hz maximum rate 00193 writeByte(MPU9250_ADDRESS, CONFIG, 0x03); 00194 00195 // Set sample rate = gyroscope output rate/(1 + SMPLRT_DIV) 00196 writeByte(MPU9250_ADDRESS, SMPLRT_DIV, 0x04); // Use a 200 Hz rate; the same rate set in CONFIG above 00197 00198 // Set gyroscope full scale range 00199 // Range selects FS_SEL and AFS_SEL are 0 - 3, so 2-bit values are left-shifted into positions 4:3 00200 uint8_t c = readByte(MPU9250_ADDRESS, GYRO_CONFIG); 00201 writeByte(MPU9250_ADDRESS, GYRO_CONFIG, c & ~0xE0); // Clear self-test bits [7:5] 00202 writeByte(MPU9250_ADDRESS, GYRO_CONFIG, c & ~0x18); // Clear AFS bits [4:3] 00203 writeByte(MPU9250_ADDRESS, GYRO_CONFIG, c | Gscale << 3); // Set full scale range for the gyro 00204 00205 // Set accelerometer configuration 00206 c = readByte(MPU9250_ADDRESS, ACCEL_CONFIG); 00207 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, c & ~0xE0); // Clear self-test bits [7:5] 00208 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, c & ~0x18); // Clear AFS bits [4:3] 00209 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, c | Ascale << 3); // Set full scale range for the accelerometer 00210 00211 // Set accelerometer sample rate configuration 00212 // It is possible to get a 4 kHz sample rate from the accelerometer by choosing 1 for 00213 // accel_fchoice_b bit [3]; in this case the bandwidth is 1.13 kHz 00214 c = readByte(MPU9250_ADDRESS, ACCEL_CONFIG2); 00215 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG2, c & ~0x0F); // Clear accel_fchoice_b (bit 3) and A_DLPFG (bits [2:0]) 00216 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG2, c | 0x03); // Set accelerometer rate to 1 kHz and bandwidth to 41 Hz 00217 00218 // The accelerometer, gyro, and thermometer are set to 1 kHz sample rates, 00219 // but all these rates are further reduced by a factor of 5 to 200 Hz because of the SMPLRT_DIV setting 00220 00221 // Configure Interrupts and Bypass Enable 00222 // Set interrupt pin active high, push-pull, and clear on read of INT_STATUS, enable I2C_BYPASS_EN so additional chips 00223 // can join the I2C bus and all can be controlled by the Arduino as master 00224 writeByte(MPU9250_ADDRESS, INT_PIN_CFG, 0x22); 00225 writeByte(MPU9250_ADDRESS, INT_ENABLE, 0x01); // Enable data ready (bit 0) interrupt 00226 } 00227 00228 // Function which accumulates gyro and accelerometer data after device initialization. It calculates the average 00229 // of the at-rest readings and then loads the resulting offsets into accelerometer and gyro bias registers. 00230 void MPU9250::calibrateMPU9250(float * dest1, float * dest2) 00231 { 00232 uint8_t data[12]; // data array to hold accelerometer and gyro x, y, z, data 00233 uint16_t ii, packet_count, fifo_count; 00234 int32_t gyro_bias[3] = {0, 0, 0}, accel_bias[3] = {0, 0, 0}; 00235 00236 // reset device, reset all registers, clear gyro and accelerometer bias registers 00237 writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x80); // Write a one to bit 7 reset bit; toggle reset device 00238 wait(0.1); 00239 00240 // get stable time source 00241 // Set clock source to be PLL with x-axis gyroscope reference, bits 2:0 = 001 00242 writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x01); 00243 writeByte(MPU9250_ADDRESS, PWR_MGMT_2, 0x00); 00244 wait(0.2); 00245 00246 // Configure device for bias calculation 00247 writeByte(MPU9250_ADDRESS, INT_ENABLE, 0x00); // Disable all interrupts 00248 writeByte(MPU9250_ADDRESS, FIFO_EN, 0x00); // Disable FIFO 00249 writeByte(MPU9250_ADDRESS, PWR_MGMT_1, 0x00); // Turn on internal clock source 00250 writeByte(MPU9250_ADDRESS, I2C_MST_CTRL, 0x00); // Disable I2C master 00251 writeByte(MPU9250_ADDRESS, USER_CTRL, 0x00); // Disable FIFO and I2C master modes 00252 writeByte(MPU9250_ADDRESS, USER_CTRL, 0x0C); // Reset FIFO and DMP 00253 wait(0.015); 00254 00255 // Configure MPU9250 gyro and accelerometer for bias calculation 00256 writeByte(MPU9250_ADDRESS, CONFIG, 0x01); // Set low-pass filter to 188 Hz 00257 writeByte(MPU9250_ADDRESS, SMPLRT_DIV, 0x00); // Set sample rate to 1 kHz 00258 writeByte(MPU9250_ADDRESS, GYRO_CONFIG, 0x00); // Set gyro full-scale to 250 degrees per second, maximum sensitivity 00259 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, 0x00); // Set accelerometer full-scale to 2 g, maximum sensitivity 00260 00261 uint16_t gyrosensitivity = 131; // = 131 LSB/degrees/sec 00262 uint16_t accelsensitivity = 16384; // = 16384 LSB/g 00263 00264 // Configure FIFO to capture accelerometer and gyro data for bias calculation 00265 writeByte(MPU9250_ADDRESS, USER_CTRL, 0x40); // Enable FIFO 00266 writeByte(MPU9250_ADDRESS, FIFO_EN, 0x78); // Enable gyro and accelerometer sensors for FIFO (max size 512 bytes in MPU-9250) 00267 wait(0.04); // accumulate 40 samples in 80 milliseconds = 480 bytes 00268 00269 // At end of sample accumulation, turn off FIFO sensor read 00270 writeByte(MPU9250_ADDRESS, FIFO_EN, 0x00); // Disable gyro and accelerometer sensors for FIFO 00271 readBytes(MPU9250_ADDRESS, FIFO_COUNTH, 2, &data[0]); // read FIFO sample count 00272 fifo_count = ((uint16_t)data[0] << 8) | data[1]; 00273 packet_count = fifo_count/12;// How many sets of full gyro and accelerometer data for averaging 00274 00275 for (ii = 0; ii < packet_count; ii++) { 00276 int16_t accel_temp[3] = {0, 0, 0}, gyro_temp[3] = {0, 0, 0}; 00277 readBytes(MPU9250_ADDRESS, FIFO_R_W, 12, &data[0]); // read data for averaging 00278 accel_temp[0] = (int16_t) (((int16_t)data[0] << 8) | data[1] ) ; // Form signed 16-bit integer for each sample in FIFO 00279 accel_temp[1] = (int16_t) (((int16_t)data[2] << 8) | data[3] ) ; 00280 accel_temp[2] = (int16_t) (((int16_t)data[4] << 8) | data[5] ) ; 00281 gyro_temp[0] = (int16_t) (((int16_t)data[6] << 8) | data[7] ) ; 00282 gyro_temp[1] = (int16_t) (((int16_t)data[8] << 8) | data[9] ) ; 00283 gyro_temp[2] = (int16_t) (((int16_t)data[10] << 8) | data[11]) ; 00284 00285 accel_bias[0] += (int32_t) accel_temp[0]; // Sum individual signed 16-bit biases to get accumulated signed 32-bit biases 00286 accel_bias[1] += (int32_t) accel_temp[1]; 00287 accel_bias[2] += (int32_t) accel_temp[2]; 00288 gyro_bias[0] += (int32_t) gyro_temp[0]; 00289 gyro_bias[1] += (int32_t) gyro_temp[1]; 00290 gyro_bias[2] += (int32_t) gyro_temp[2]; 00291 00292 } 00293 accel_bias[0] /= (int32_t) packet_count; // Normalize sums to get average count biases 00294 accel_bias[1] /= (int32_t) packet_count; 00295 accel_bias[2] /= (int32_t) packet_count; 00296 gyro_bias[0] /= (int32_t) packet_count; 00297 gyro_bias[1] /= (int32_t) packet_count; 00298 gyro_bias[2] /= (int32_t) packet_count; 00299 00300 if(accel_bias[2] > 0L) {accel_bias[2] -= (int32_t) accelsensitivity;} // Remove gravity from the z-axis accelerometer bias calculation 00301 else {accel_bias[2] += (int32_t) accelsensitivity;} 00302 00303 // Construct the gyro biases for push to the hardware gyro bias registers, which are reset to zero upon device startup 00304 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 00305 data[1] = (-gyro_bias[0]/4) & 0xFF; // Biases are additive, so change sign on calculated average gyro biases 00306 data[2] = (-gyro_bias[1]/4 >> 8) & 0xFF; 00307 data[3] = (-gyro_bias[1]/4) & 0xFF; 00308 data[4] = (-gyro_bias[2]/4 >> 8) & 0xFF; 00309 data[5] = (-gyro_bias[2]/4) & 0xFF; 00310 00311 /// Push gyro biases to hardware registers 00312 /* writeByte(MPU9250_ADDRESS, XG_OFFSET_H, data[0]); 00313 writeByte(MPU9250_ADDRESS, XG_OFFSET_L, data[1]); 00314 writeByte(MPU9250_ADDRESS, YG_OFFSET_H, data[2]); 00315 writeByte(MPU9250_ADDRESS, YG_OFFSET_L, data[3]); 00316 writeByte(MPU9250_ADDRESS, ZG_OFFSET_H, data[4]); 00317 writeByte(MPU9250_ADDRESS, ZG_OFFSET_L, data[5]); 00318 */ 00319 dest1[0] = (float) gyro_bias[0]/(float) gyrosensitivity; // construct gyro bias in deg/s for later manual subtraction 00320 dest1[1] = (float) gyro_bias[1]/(float) gyrosensitivity; 00321 dest1[2] = (float) gyro_bias[2]/(float) gyrosensitivity; 00322 00323 // Construct the accelerometer biases for push to the hardware accelerometer bias registers. These registers contain 00324 // factory trim values which must be added to the calculated accelerometer biases; on boot up these registers will hold 00325 // non-zero values. In addition, bit 0 of the lower byte must be preserved since it is used for temperature 00326 // compensation calculations. Accelerometer bias registers expect bias input as 2048 LSB per g, so that 00327 // the accelerometer biases calculated above must be divided by 8. 00328 00329 int32_t accel_bias_reg[3] = {0, 0, 0}; // A place to hold the factory accelerometer trim biases 00330 readBytes(MPU9250_ADDRESS, XA_OFFSET_H, 2, &data[0]); // Read factory accelerometer trim values 00331 accel_bias_reg[0] = (int16_t) ((int16_t)data[0] << 8) | data[1]; 00332 readBytes(MPU9250_ADDRESS, YA_OFFSET_H, 2, &data[0]); 00333 accel_bias_reg[1] = (int16_t) ((int16_t)data[0] << 8) | data[1]; 00334 readBytes(MPU9250_ADDRESS, ZA_OFFSET_H, 2, &data[0]); 00335 accel_bias_reg[2] = (int16_t) ((int16_t)data[0] << 8) | data[1]; 00336 00337 uint32_t mask = 1uL; // Define mask for temperature compensation bit 0 of lower byte of accelerometer bias registers 00338 uint8_t mask_bit[3] = {0, 0, 0}; // Define array to hold mask bit for each accelerometer bias axis 00339 00340 for(ii = 0; ii < 3; ii++) { 00341 if(accel_bias_reg[ii] & mask) mask_bit[ii] = 0x01; // If temperature compensation bit is set, record that fact in mask_bit 00342 } 00343 00344 // Construct total accelerometer bias, including calculated average accelerometer bias from above 00345 accel_bias_reg[0] -= (accel_bias[0]/8); // Subtract calculated averaged accelerometer bias scaled to 2048 LSB/g (16 g full scale) 00346 accel_bias_reg[1] -= (accel_bias[1]/8); 00347 accel_bias_reg[2] -= (accel_bias[2]/8); 00348 00349 data[0] = (accel_bias_reg[0] >> 8) & 0xFF; 00350 data[1] = (accel_bias_reg[0]) & 0xFF; 00351 data[1] = data[1] | mask_bit[0]; // preserve temperature compensation bit when writing back to accelerometer bias registers 00352 data[2] = (accel_bias_reg[1] >> 8) & 0xFF; 00353 data[3] = (accel_bias_reg[1]) & 0xFF; 00354 data[3] = data[3] | mask_bit[1]; // preserve temperature compensation bit when writing back to accelerometer bias registers 00355 data[4] = (accel_bias_reg[2] >> 8) & 0xFF; 00356 data[5] = (accel_bias_reg[2]) & 0xFF; 00357 data[5] = data[5] | mask_bit[2]; // preserve temperature compensation bit when writing back to accelerometer bias registers 00358 00359 // Apparently this is not working for the acceleration biases in the MPU-9250 00360 // Are we handling the temperature correction bit properly? 00361 // Push accelerometer biases to hardware registers 00362 /* writeByte(MPU9250_ADDRESS, XA_OFFSET_H, data[0]); 00363 writeByte(MPU9250_ADDRESS, XA_OFFSET_L, data[1]); 00364 writeByte(MPU9250_ADDRESS, YA_OFFSET_H, data[2]); 00365 writeByte(MPU9250_ADDRESS, YA_OFFSET_L, data[3]); 00366 writeByte(MPU9250_ADDRESS, ZA_OFFSET_H, data[4]); 00367 writeByte(MPU9250_ADDRESS, ZA_OFFSET_L, data[5]); 00368 */ 00369 // Output scaled accelerometer biases for manual subtraction in the main program 00370 dest2[0] = (float)accel_bias[0]/(float)accelsensitivity; 00371 dest2[1] = (float)accel_bias[1]/(float)accelsensitivity; 00372 dest2[2] = (float)accel_bias[2]/(float)accelsensitivity; 00373 } 00374 00375 00376 // Accelerometer and gyroscope self test; check calibration wrt factory settings 00377 void MPU9250::MPU9250SelfTest(float * destination) // Should return percent deviation from factory trim values, +/- 14 or less deviation is a pass 00378 { 00379 uint8_t rawData[6] = {0, 0, 0, 0, 0, 0}; 00380 uint8_t selfTest[6]; 00381 int16_t gAvg[3], aAvg[3], aSTAvg[3], gSTAvg[3]; 00382 float factoryTrim[6]; 00383 uint8_t FS = 0; 00384 00385 writeByte(MPU9250_ADDRESS, SMPLRT_DIV, 0x00); // Set gyro sample rate to 1 kHz 00386 writeByte(MPU9250_ADDRESS, CONFIG, 0x02); // Set gyro sample rate to 1 kHz and DLPF to 92 Hz 00387 writeByte(MPU9250_ADDRESS, GYRO_CONFIG, 1<<FS); // Set full scale range for the gyro to 250 dps 00388 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG2, 0x02); // Set accelerometer rate to 1 kHz and bandwidth to 92 Hz 00389 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, 1<<FS); // Set full scale range for the accelerometer to 2 g 00390 00391 for( int ii = 0; ii < 200; ii++) { // get average current values of gyro and acclerometer 00392 00393 readBytes(MPU9250_ADDRESS, ACCEL_XOUT_H, 6, &rawData[0]); // Read the six raw data registers into data array 00394 aAvg[0] += (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value 00395 aAvg[1] += (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; 00396 aAvg[2] += (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; 00397 00398 readBytes(MPU9250_ADDRESS, GYRO_XOUT_H, 6, &rawData[0]); // Read the six raw data registers sequentially into data array 00399 gAvg[0] += (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value 00400 gAvg[1] += (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; 00401 gAvg[2] += (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; 00402 } 00403 00404 for (int ii =0; ii < 3; ii++) { // Get average of 200 values and store as average current readings 00405 aAvg[ii] /= 200; 00406 gAvg[ii] /= 200; 00407 } 00408 00409 // Configure the accelerometer for self-test 00410 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, 0xE0); // Enable self test on all three axes and set accelerometer range to +/- 2 g 00411 writeByte(MPU9250_ADDRESS, GYRO_CONFIG, 0xE0); // Enable self test on all three axes and set gyro range to +/- 250 degrees/s 00412 wait_ms(25); // Delay a while to let the device stabilize 00413 00414 for( int ii = 0; ii < 200; ii++) { // get average self-test values of gyro and acclerometer 00415 00416 readBytes(MPU9250_ADDRESS, ACCEL_XOUT_H, 6, &rawData[0]); // Read the six raw data registers into data array 00417 aSTAvg[0] += (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value 00418 aSTAvg[1] += (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; 00419 aSTAvg[2] += (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; 00420 00421 readBytes(MPU9250_ADDRESS, GYRO_XOUT_H, 6, &rawData[0]); // Read the six raw data registers sequentially into data array 00422 gSTAvg[0] += (int16_t)(((int16_t)rawData[0] << 8) | rawData[1]) ; // Turn the MSB and LSB into a signed 16-bit value 00423 gSTAvg[1] += (int16_t)(((int16_t)rawData[2] << 8) | rawData[3]) ; 00424 gSTAvg[2] += (int16_t)(((int16_t)rawData[4] << 8) | rawData[5]) ; 00425 } 00426 00427 for (int ii =0; ii < 3; ii++) { // Get average of 200 values and store as average self-test readings 00428 aSTAvg[ii] /= 200; 00429 gSTAvg[ii] /= 200; 00430 } 00431 00432 // Configure the gyro and accelerometer for normal operation 00433 writeByte(MPU9250_ADDRESS, ACCEL_CONFIG, 0x00); 00434 writeByte(MPU9250_ADDRESS, GYRO_CONFIG, 0x00); 00435 wait_ms(25); // Delay a while to let the device stabilize 00436 00437 // Retrieve accelerometer and gyro factory Self-Test Code from USR_Reg 00438 selfTest[0] = readByte(MPU9250_ADDRESS, SELF_TEST_X_ACCEL); // X-axis accel self-test results 00439 selfTest[1] = readByte(MPU9250_ADDRESS, SELF_TEST_Y_ACCEL); // Y-axis accel self-test results 00440 selfTest[2] = readByte(MPU9250_ADDRESS, SELF_TEST_Z_ACCEL); // Z-axis accel self-test results 00441 selfTest[3] = readByte(MPU9250_ADDRESS, SELF_TEST_X_GYRO); // X-axis gyro self-test results 00442 selfTest[4] = readByte(MPU9250_ADDRESS, SELF_TEST_Y_GYRO); // Y-axis gyro self-test results 00443 selfTest[5] = readByte(MPU9250_ADDRESS, SELF_TEST_Z_GYRO); // Z-axis gyro self-test results 00444 00445 // Retrieve factory self-test value from self-test code reads 00446 factoryTrim[0] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[0] - 1.0) )); // FT[Xa] factory trim calculation 00447 factoryTrim[1] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[1] - 1.0) )); // FT[Ya] factory trim calculation 00448 factoryTrim[2] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[2] - 1.0) )); // FT[Za] factory trim calculation 00449 factoryTrim[3] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[3] - 1.0) )); // FT[Xg] factory trim calculation 00450 factoryTrim[4] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[4] - 1.0) )); // FT[Yg] factory trim calculation 00451 factoryTrim[5] = (float)(2620/1<<FS)*(pow( 1.01 , ((float)selfTest[5] - 1.0) )); // FT[Zg] factory trim calculation 00452 00453 // Report results as a ratio of (STR - FT)/FT; the change from Factory Trim of the Self-Test Response 00454 // To get percent, must multiply by 100 00455 for (int i = 0; i < 3; i++) { 00456 destination[i] = 100.0*((float)(aSTAvg[i] - aAvg[i]))/factoryTrim[i]; // Report percent differences 00457 destination[i+3] = 100.0*((float)(gSTAvg[i] - gAvg[i]))/factoryTrim[i+3]; // Report percent differences 00458 } 00459 00460 } 00461 00462 00463 00464 // Implementation of Sebastian Madgwick's "...efficient orientation filter for... inertial/magnetic sensor arrays" 00465 // (see http://www.x-io.co.uk/category/open-source/ for examples and more details) 00466 // which fuses acceleration, rotation rate, and magnetic moments to produce a quaternion-based estimate of absolute 00467 // device orientation -- which can be converted to yaw, pitch, and roll. Useful for stabilizing quadcopters, etc. 00468 // The performance of the orientation filter is at least as good as conventional Kalman-based filtering algorithms 00469 // but is much less computationally intensive---it can be performed on a 3.3 V Pro Mini operating at 8 MHz! 00470 void MPU9250::MadgwickQuaternionUpdate(float ax, float ay, float az, float gx, float gy, float gz, float mx, float my, float mz) 00471 { 00472 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 00473 float beta = sqrt(3.0f / 4.0f) * GyroMeasError; // compute beta 00474 float GyroMeasDrift = PI * (1.0f / 180.0f); // gyroscope measurement drift in rad/s/s (start at 0.0 deg/s/s) 00475 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 00476 q[0] = 1.0f; 00477 q[1] = 0.0f; 00478 q[2] = 0.0f; 00479 q[3] = 0.0f; 00480 float q1 = q[0], q2 = q[1], q3 = q[2], q4 = q[3]; // short name local variable for readability 00481 float norm; 00482 float hx, hy, _2bx, _2bz; 00483 float s1, s2, s3, s4; 00484 float qDot1, qDot2, qDot3, qDot4; 00485 00486 // Auxiliary variables to avoid repeated arithmetic 00487 float _2q1mx; 00488 float _2q1my; 00489 float _2q1mz; 00490 float _2q2mx; 00491 float _4bx; 00492 float _4bz; 00493 float _2q1 = 2.0f * q1; 00494 float _2q2 = 2.0f * q2; 00495 float _2q3 = 2.0f * q3; 00496 float _2q4 = 2.0f * q4; 00497 float _2q1q3 = 2.0f * q1 * q3; 00498 float _2q3q4 = 2.0f * q3 * q4; 00499 float q1q1 = q1 * q1; 00500 float q1q2 = q1 * q2; 00501 float q1q3 = q1 * q3; 00502 float q1q4 = q1 * q4; 00503 float q2q2 = q2 * q2; 00504 float q2q3 = q2 * q3; 00505 float q2q4 = q2 * q4; 00506 float q3q3 = q3 * q3; 00507 float q3q4 = q3 * q4; 00508 float q4q4 = q4 * q4; 00509 00510 // Normalise accelerometer measurement 00511 norm = sqrt(ax * ax + ay * ay + az * az); 00512 if (norm == 0.0f) return; // handle NaN 00513 norm = 1.0f/norm; 00514 ax *= norm; 00515 ay *= norm; 00516 az *= norm; 00517 00518 // Normalise magnetometer measurement 00519 norm = sqrt(mx * mx + my * my + mz * mz); 00520 if (norm == 0.0f) return; // handle NaN 00521 norm = 1.0f/norm; 00522 mx *= norm; 00523 my *= norm; 00524 mz *= norm; 00525 00526 // Reference direction of Earth's magnetic field 00527 _2q1mx = 2.0f * q1 * mx; 00528 _2q1my = 2.0f * q1 * my; 00529 _2q1mz = 2.0f * q1 * mz; 00530 _2q2mx = 2.0f * q2 * mx; 00531 hx = mx * q1q1 - _2q1my * q4 + _2q1mz * q3 + mx * q2q2 + _2q2 * my * q3 + _2q2 * mz * q4 - mx * q3q3 - mx * q4q4; 00532 hy = _2q1mx * q4 + my * q1q1 - _2q1mz * q2 + _2q2mx * q3 - my * q2q2 + my * q3q3 + _2q3 * mz * q4 - my * q4q4; 00533 _2bx = sqrt(hx * hx + hy * hy); 00534 _2bz = -_2q1mx * q3 + _2q1my * q2 + mz * q1q1 + _2q2mx * q4 - mz * q2q2 + _2q3 * my * q4 - mz * q3q3 + mz * q4q4; 00535 _4bx = 2.0f * _2bx; 00536 _4bz = 2.0f * _2bz; 00537 00538 // Gradient decent algorithm corrective step 00539 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); 00540 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); 00541 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); 00542 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); 00543 norm = sqrt(s1 * s1 + s2 * s2 + s3 * s3 + s4 * s4); // normalise step magnitude 00544 norm = 1.0f/norm; 00545 s1 *= norm; 00546 s2 *= norm; 00547 s3 *= norm; 00548 s4 *= norm; 00549 00550 // Compute rate of change of quaternion 00551 qDot1 = 0.5f * (-q2 * gx - q3 * gy - q4 * gz) - beta * s1; 00552 qDot2 = 0.5f * (q1 * gx + q3 * gz - q4 * gy) - beta * s2; 00553 qDot3 = 0.5f * (q1 * gy - q2 * gz + q4 * gx) - beta * s3; 00554 qDot4 = 0.5f * (q1 * gz + q2 * gy - q3 * gx) - beta * s4; 00555 00556 // Integrate to yield quaternion 00557 q1 += qDot1 * deltat; 00558 q2 += qDot2 * deltat; 00559 q3 += qDot3 * deltat; 00560 q4 += qDot4 * deltat; 00561 norm = sqrt(q1 * q1 + q2 * q2 + q3 * q3 + q4 * q4); // normalise quaternion 00562 norm = 1.0f/norm; 00563 q[0] = q1 * norm; 00564 q[1] = q2 * norm; 00565 q[2] = q3 * norm; 00566 q[3] = q4 * norm; 00567 00568 } 00569 00570 00571 00572 // Similar to Madgwick scheme but uses proportional and integral filtering on the error between estimated reference vectors and 00573 // measured ones. 00574 void MPU9250::MahonyQuaternionUpdate(float ax, float ay, float az, float gx, float gy, float gz, float mx, float my, float mz) 00575 { 00576 float eInt[3] = {0.0f, 0.0f, 0.0f}; // vector to hold integral error for Mahony method 00577 // q[0] = 1.0f; 00578 // q[1] = 0.0f; 00579 // q[2] = 0.0f; 00580 // q[3] = 0.0f; 00581 float q1 = q[0], q2 = q[1], q3 = q[2], q4 = q[3]; // short name local variable for readability 00582 float norm; 00583 float hx, hy, bx, bz; 00584 float vx, vy, vz, wx, wy, wz; 00585 float ex, ey, ez; 00586 float pa, pb, pc; 00587 00588 // Auxiliary variables to avoid repeated arithmetic 00589 float q1q1 = q1 * q1; 00590 float q1q2 = q1 * q2; 00591 float q1q3 = q1 * q3; 00592 float q1q4 = q1 * q4; 00593 float q2q2 = q2 * q2; 00594 float q2q3 = q2 * q3; 00595 float q2q4 = q2 * q4; 00596 float q3q3 = q3 * q3; 00597 float q3q4 = q3 * q4; 00598 float q4q4 = q4 * q4; 00599 00600 // Normalise accelerometer measurement 00601 norm = sqrt(ax * ax + ay * ay + az * az); 00602 if (norm == 0.0f) return; // handle NaN 00603 norm = 1.0f / norm; // use reciprocal for division 00604 ax *= norm; 00605 ay *= norm; 00606 az *= norm; 00607 00608 // Normalise magnetometer measurement 00609 norm = sqrt(mx * mx + my * my + mz * mz); 00610 if (norm == 0.0f) return; // handle NaN 00611 norm = 1.0f / norm; // use reciprocal for division 00612 mx *= norm; 00613 my *= norm; 00614 mz *= norm; 00615 00616 // Reference direction of Earth's magnetic field 00617 hx = 2.0f * mx * (0.5f - q3q3 - q4q4) + 2.0f * my * (q2q3 - q1q4) + 2.0f * mz * (q2q4 + q1q3); 00618 hy = 2.0f * mx * (q2q3 + q1q4) + 2.0f * my * (0.5f - q2q2 - q4q4) + 2.0f * mz * (q3q4 - q1q2); 00619 bx = sqrt((hx * hx) + (hy * hy)); 00620 bz = 2.0f * mx * (q2q4 - q1q3) + 2.0f * my * (q3q4 + q1q2) + 2.0f * mz * (0.5f - q2q2 - q3q3); 00621 00622 // Estimated direction of gravity and magnetic field 00623 vx = 2.0f * (q2q4 - q1q3); 00624 vy = 2.0f * (q1q2 + q3q4); 00625 vz = q1q1 - q2q2 - q3q3 + q4q4; 00626 wx = 2.0f * bx * (0.5f - q3q3 - q4q4) + 2.0f * bz * (q2q4 - q1q3); 00627 wy = 2.0f * bx * (q2q3 - q1q4) + 2.0f * bz * (q1q2 + q3q4); 00628 wz = 2.0f * bx * (q1q3 + q2q4) + 2.0f * bz * (0.5f - q2q2 - q3q3); 00629 00630 // Error is cross product between estimated direction and measured direction of gravity 00631 ex = (ay * vz - az * vy) + (my * wz - mz * wy); 00632 ey = (az * vx - ax * vz) + (mz * wx - mx * wz); 00633 ez = (ax * vy - ay * vx) + (mx * wy - my * wx); 00634 if (Ki > 0.0f) 00635 { 00636 eInt[0] += ex; // accumulate integral error 00637 eInt[1] += ey; 00638 eInt[2] += ez; 00639 } 00640 else 00641 { 00642 eInt[0] = 0.0f; // prevent integral wind up 00643 eInt[1] = 0.0f; 00644 eInt[2] = 0.0f; 00645 } 00646 00647 // Apply feedback terms 00648 gx = gx + Kp * ex + Ki * eInt[0]; 00649 gy = gy + Kp * ey + Ki * eInt[1]; 00650 gz = gz + Kp * ez + Ki * eInt[2]; 00651 00652 // Integrate rate of change of quaternion 00653 pa = q2; 00654 pb = q3; 00655 pc = q4; 00656 q1 = q1 + (-q2 * gx - q3 * gy - q4 * gz) * (0.5f * deltat); 00657 q2 = pa + (q1 * gx + pb * gz - pc * gy) * (0.5f * deltat); 00658 q3 = pb + (q1 * gy - pa * gz + pc * gx) * (0.5f * deltat); 00659 q4 = pc + (q1 * gz + pa * gy - pb * gx) * (0.5f * deltat); 00660 00661 // Normalise quaternion 00662 norm = sqrt(q1 * q1 + q2 * q2 + q3 * q3 + q4 * q4); 00663 norm = 1.0f / norm; 00664 q[0] = q1 * norm; 00665 q[1] = q2 * norm; 00666 q[2] = q3 * norm; 00667 q[3] = q4 * norm; 00668 00669 }
Generated on Thu Aug 11 2022 19:32:11 by
1.7.2