MEMS IMU Sensor Orientation Algorithms-Comparison in a Simulation Environment

Accurate and precise estimation of orientation from low cost inertial measurement units (IMUs) is a challenge due to the influence of noises from the accelerometer, gyroscope and magnetometer. For practical applications, it would be useful to first quantify the expected error in estimated orientation for a known noise associated with a sensor so as to make an optimal selection of a sensor and fusion algorithm based on application requirement. There are a wide range of sensor fusion algorithms in literature to make these angular measurements from MEMS based IMUs. This paper proposes use of a simulation platform for comparative performance assessment of orientation algorithms for 9 axis IMUs in presence of internal noises and demonstrates with examples the benefits of the same. The simulated accelerometer, gyroscope and magnetometer signals in the IMU are perturbed with white Gaussian noise and bias instability. The effects of these noisy inputs on the orientation output of algorithms are calculated using MATLAB, under static as well as three types of dynamic motion profiles. The best choice of algorithm is determined by comparing the mean value and standard deviation of the shortest angle measure between the true orientation and estimated orientation from multiple simulations for each of the algorithms under evaluation. Among the four orientation algorithms compared here, the Mahony orientation algorithm produced good results for all simulations of static and dynamic motion profiles with an error less than 8° for the sensor modeled in simulation platform over the duration of the simulation.