EE 440 Final Presentation Compass and Gyro Integration for 1D Case by Olson, David

1. EE 440 Final PresentationCompass and Gyro Integration for 1D Caseby Olson, David EE 440 Compass and Gyro integration for 1D case

2. Presentation Outline • Objective • Introduction and Background • Method and Approach • Discussion of Results • Conclusion and Summary • Questions EE 440 Compass and Gyro integration for 1D case

3. Objective • Problem Statement • Determine methods of integration of a level-compass with z-gyro to determine heading of a single body. Heading from Compass ? Improved Heading Heading from Z-Gyro EE 440 Compass and Gyro integration for 1D case

4. Objective Cont. • Methods of Sensor Integration • Unweighted Average • Weighted Average • Complimentary Filter • Kalman Filter ? Heading from Compass Improved Heading Heading from Z-Gyro EE 440 Compass and Gyro integration for 1D case

5. Introduction and Background • Application • Integration of these sensors can lead to reliable systems such as AHRS and aiming weapons systems. • Relevance • Having reliable data provides necessary performance for such critical systems in regards to AHRS and weapons systems. • Sensor Methods • Compass can provide heading with measurements of magnetic fields • Gyroscopes provides angular rate, which can be integrated to provide angular position EE 440 Compass and Gyro integration for 1D case

6. Introduction and Background Cont. • Gyro Calibration • Before Calibration • After Calibration The gyro data is then integrated, with an initial condition of the first compass reading to initialize gyro heading. EE 440 Compass and Gyro integration for 1D case

7. Introduction and Background Cont. • Compass Calibration • Compass hard iron calibrations are taken into account. • X & Y magnetometer readings are converted into heading. • Data Calibration ? EE 440 Compass and Gyro integration for 1D case

8. Method and Approach Cont. • Unweighted Average Unweighted Average EE 440 Compass and Gyro integration for 1D case

9. Method and Approach Cont. • Weighted Average • The average is weighted and scaled according to the variance in each signal. • Compass variance is taken from calibration data where the compass sits still • Gyro variance is taken from the first two seconds of data before the VN-200 begins moving EE 440 Compass and Gyro integration for 1D case

10. Method and Approach Cont. • Complimentary Filter • Below is an example of a complimentary filter. • The complimentary filter combines the low frequency content of one signal with the high frequency content of another. EE 440 Compass and Gyro integration for 1D case

11. Method and Approach Cont. • Complimentary Filter Cont. • This is the variation of the complimentary filter that was implemented, which accomplishes the same goal. EE 440 Compass and Gyro integration for 1D case

12. Method and Approach Cont. • Kalman Filter • To implement the Kalman Filter, the following method was used: • First, a dynamic model of the error must be made. EE 440 Compass and Gyro integration for 1D case

13. Method and Approach Cont. • Kalman Filter Cont. • The Overall Model looks like this: = + EE 440 Compass and Gyro integration for 1D case

14. Method and Approach Cont. • Kalman Filter Cont. EE 440 Compass and Gyro integration for 1D case

15. Method and Approach • Test Procedure Example Data EE 440 Compass and Gyro integration for 1D case

16. Method and Approach Cont. • Sensor Data Alone, Unweighted Average, Weighted Average EE 440 Compass and Gyro integration for 1D case

17. Method and Approach Cont. • Complimentary Filter, Kalman Filter EE 440 Compass and Gyro integration for 1D case

18. Method and Approach Cont. • Sensor Data Alone, Unweighted Average, Weighted Average EE 440 Compass and Gyro integration for 1D case

19. Method and Approach Cont. • Complimentary Filter, Kalman Filter EE 440 Compass and Gyro integration for 1D case

20. Discussion of Results • Comparison of all four methods: • Gyro by itself has to much drift to be reliable. • Compass matches end to end yet has too much of a scale factor issue, most likely a soft iron calibration issue. • Unweighted Average is noisy but splits the two individual measurements well. • Weighted Average fits the data better but relied too heavy on our compass for the weighting values used. • Complimentary Filter works well, but is still noisy and does not match well to ending data. • Kalman filter had the best performance, but still has not reached its true potential. EE 440 Compass and Gyro integration for 1D case

21. Conclusion and Summary • Conclusion • There are multiple methods for integrating an aiding sensor with INS measurement systems. • Some methods work better than others, but require more complexity, the method chosen should match the performance needed by the system. • Unweighted Average • Weighted Average • Complimentary Filter • Kalman Filter EE 440 Compass and Gyro integration for 1D case

22. Conclusion and Summary Cont. • Summary There are many methods to integrate the sensors, pick which method best meets the needs of your project. Heading from Compass Unweighted Average Weighted Average Complimentary Filter Kalman Filter Improved Heading Heading from Z-Gyro EE 440 Compass and Gyro integration for 1D case

23. Conclusion and Summary Cont. • Things I would do differently if I did this project again • I would try to provide quantitative comparisons for each method of sensor integration. • I would be more thorough with both gyro and compass calibrations. • I would take more data to spin around an entire circle as opposed to one fourth of the circle. EE 440 Compass and Gyro integration for 1D case

24. Questions Questions? David Olson 512.496.7290 olsond3@my.erau.edu EE 440 Compass and Gyro integration for 1D case