원문정보
초록
영어
Attitude estimation has a wide range of applications including aerial (UAVs for example), underwater (ROVs for instance), navigation systems, robotics, games, augmented reality system, industrial and so on. Extensive research over decades in this field resulted in a number of powerful estimators; complex like Kalman based algorithms as well as simple such as complementary filters & the kinds. For applications where computational simplicity is of prime concern, complementary filters have proven efficiency. This paper presents a comparative study of computationally simple algorithms naming Explicit Complementary Filter (ECF) and Gradient Descent based Complementary Filter (GDCF) along with computationally demanding extended Kalman filter-a de-facto standard for attitude estimation so far, regarding attitude estimation based on MEMS IMU. An alternative would be the variants of complementary filter; sufficiently efficient and simple scheme to avoid computational complexity.Performance of these estimators is evaluated for Euler angle estimation using both simulated data from Matlab and experimental data from MPU6050 IMU. The assessment is based on the root mean square error computation for these algorithms. Moreover, the algorithms adjustable parameters were exploited for a range of values in the hunt for perfection
목차
1. Introduction
2. Mems IMU
2.1. Accelerometer Model
2.2. Gyroscope Model
3. Estimators/Algorithms
3.1. Extended Kalman Filter
3.2. Explicit Complementary Filter
3.3. Gradient Decent Based Complementary Filter
4. Simulation Results
5. Experimental Results
6. Conclusion
References