'각도'에 해당되는 글 1건

  1. 2017.11.29 MPU6050의 칼만 필터(Kalman filter)의 구현 예제(4)


아두이노(Arduino) 환경에서 전형적인 MPU6050 센서 입력에 사용되는 칼만 필터(Kalman filter)의 예제입니다.

http://blog.tkjelectronics.dk/2012/09/a-practical-approach-to-kalman-filter-and-how-to-implement-it/#comment-57783


또한 이 칼만 필터는 가속도계(accelerometer) 혹은 지자계(magnetometer)와 그리고 자이로(gyroscope)로부터 각도(angle), 각속도(rate) 그리고 bias를 계산하는데 사용될 수 있습니다(C++ version).

https://github.com/TKJElectronics/KalmanFilter


다음은 C version입니다.

http://www.cnblogs.com/zjutlitao/p/3915786.html



Kalman.h

/* Copyright (C) 2012 Kristian Lauszus, TKJ Electronics-> All rights reserved->


 This software may be distributed and modified under the terms of the GNU

 General Public License version 2 (GPL2) as published by the Free Software

 Foundation and appearing in the file GPL2->TXT included in the packaging of

 this file-> Please note that GPL2 Section 2[b] requires that all works based

 on this software must also be made publicly available under the terms of

 the GPL2 ("Copyleft")->


 Contact information

 -------------------


 Kristian Lauszus, TKJ Electronics

 Web      :  http://www->tkjelectronics->com

 e-mail   :  kristianl@tkjelectronics->com

 */


#ifndef _Kalman_h

#define _Kalman_h

struct Kalman {

    /* Kalman filter variables */

    double Q_angle; // Process noise variance for the accelerometer

    double Q_bias; // Process noise variance for the gyro bias

    double R_measure; // Measurement noise variance - this is actually the variance of the measurement noise


    double angle; // The angle calculated by the Kalman filter - part of the 2x1 state vector

    double bias; // The gyro bias calculated by the Kalman filter - part of the 2x1 state vector

    double rate; // Unbiased rate calculated from the rate and the calculated bias - you have to call getAngle to update the rate


    double P[2][2]; // Error covariance matrix - This is a 2x2 matrix

    double K[2]; // Kalman gain - This is a 2x1 vector

    double y; // Angle difference

    double S; // Estimate error

};


void Init(struct Kalman* klm){

    /* We will set the variables like so, these can also be tuned by the user */

    klm->Q_angle = 0.001;

    klm->Q_bias = 0.003;

    klm->R_measure = 0.03;


    klm->angle = 0; // Reset the angle

    klm->bias = 0; // Reset bias

   klm->P[0][0] = 0; // Since we assume that the bias is 0 and we know the starting angle (use setAngle), the error covariance matrix is set like so - see: http://en->wikipedia->org/wiki/Kalman_filter#Example_application->2C_technical

    klm->P[0][1] = 0;

    klm->P[1][0] = 0;

    klm->P[1][1] = 0;

}


// The angle should be in degrees and the rate should be in degrees per second and the delta time in seconds

double getAngle(struct Kalman * klm, double newAngle, double newRate, double dt) {

    // KasBot V2  -  Kalman filter module - http://www->x-firm->com/?page_id=145

    // Modified by Kristian Lauszus

   // See my blog post for more information: http://blog->tkjelectronics->dk/2012/09/a-practical-approach-to-kalman-filter-and-how-to-implement-it


    float P00_temp;

    float P01_temp;


    // Discrete Kalman filter time update equations - Time Update ("Predict")

    // Update xhat - Project the state ahead

    /* Step 1 */

    klm->rate = newRate - klm->bias;

    klm->angle += dt * klm->rate;

    

    // Update estimation error covariance - Project the error covariance ahead

    /* Step 2 */

    klm->P[0][0] += dt * (dt*klm->P[1][1] - klm->P[0][1] - klm->P[1][0] + klm->Q_angle);

    klm->P[0][1] -= dt * klm->P[1][1];

    klm->P[1][0] -= dt * klm->P[1][1];

    klm->P[1][1] += klm->Q_bias * dt;

    

    // Discrete Kalman filter measurement update equations - Measurement Update ("Correct")

    // Calculate Kalman gain - Compute the Kalman gain

    /* Step 4 */

    klm->S = klm->P[0][0] + klm->R_measure;


    /* Step 5 */

    klm->K[0] = klm->P[0][0] / klm->S;

    klm->K[1] = klm->P[1][0] / klm->S;

    

    // Calculate angle and bias - Update estimate with measurement zk (newAngle)

    /* Step 3 */

    klm->y = newAngle - klm->angle;


    /* Step 6 */

    klm->angle += klm->K[0] * klm->y;

    klm->bias += klm->K[1] * klm->y;

    

    // Calculate estimation error covariance - Update the error covariance

    /* Step 7 */

   P00_temp = klm->P[0][0];

   P01_temp = klm->P[0][1];


    klm->P[0][0] -= klm->K[0] * P00_temp;

    klm->P[0][1] -= klm->K[0] * P01_temp;

    klm->P[1][0] -= klm->K[1] * P00_temp;

    klm->P[1][1] -= klm->K[1] * P01_temp;


    return klm->angle;

}


 // Used to set angle, this should be set as the starting angle

void setAngle(struct Kalman* klm, double newAngle) { klm->angle = newAngle; }

// Return the unbiased rate

double getRate(struct Kalman* klm) { return klm->rate; }
 

/* These are used to tune the Kalman filter */

void setQangle(struct Kalman* klm, double newQ_angle) { klm->Q_angle = newQ_angle; }


/* Default value is (0.003f), raise this to follow input more closely, lower this to smooth result of kalman filter */

void setQbias(struct Kalman* klm, double newQ_bias) { klm->Q_bias = newQ_bias; }


void setRmeasure(struct Kalman* klm, double newR_measure) { klm->R_measure = newR_measure; }

double getQangle(struct Kalman* klm) { return klm->Q_angle; }

double getQbias(struct Kalman* klm) { return klm->Q_bias; }

double getRmeasure(struct Kalman* klm) { return klm->R_measure; }


#endif


Kalman.h



Posted by Nature & Life