Magnetometer-Augmented IMU Simulator: In-Depth Elaboration

Abstract

The location of objects is a growing research topic due, for instance, to the expansion of civil drones or intelligent vehicles. This expansion was made possible through the development of microelectromechanical systems (MEMS), inexpensive and miniaturized inertial sensors. In this context, this article describes the development of a new simulator which generates sensor measurements, giving a specific input trajectory. This will allow the comparison of pose estimation algorithms. To develop this simulator, the measurement equations of every type of sensor have to be analytically determined. To achieve this objective, classical kinematic equations are used for the more common sensors, i.e., accelerometers and rate gyroscopes. As nowadays, the MEMS inertial measurement units (IMUs) are generally magnetometer-augmented, an absolute world magnetic model is implemented. After the determination of the perfect measurement (through the error-free sensor models), realistic error models are developed to simulate real IMU behavior. Finally, the developed simulator is subjected to different validation tests.

Estimation of Gait Mechanics Based on Simulated and Measured IMU Data Using an Artificial Neural Network

“…The correlation found by Young et al 2011, Brunner et al (2015) and Zimmermann et al (2018), who proposed frameworks for simulating IMU data, is comparable to the results presented in this study. Young et al (2011) and Brunner et al (2015) tested their simulator for leg swinging and single rigid body movements achieving very good correlation coefficients (r >0.97). Since during this motion no impact occurs which causes soft tissue movements the results are better than the ones achieved in this study.…”

“…These aredue to the novelty of the system -not openly available from IMU sensors. To overcome the lack of a large amount of data, their synthesizing is one reasonable solution (Young et al, 2011;Brunner et al, 2015;Zimmermann et al, 2018;Mundt et al, 2019c). Young et al (2011) was the first who simulated IMU data from existing optical motion capture data to enlarge a dataset for pose estimation.…”

“…Young et al (2011) was the first who simulated IMU data from existing optical motion capture data to enlarge a dataset for pose estimation. This approach was taken a step further and validated by Brunner et al (2015) and Zimmermann et al (2018). In previous work, we simulated IMU data from optical data as well, but only validated the simulation based on a single participant (Mundt et al, 2019c).…”

Untitled

“…The sensor band we used more recently 2 returns acceleration in units of g, rotational velocity in degrees per second, and orientation angles in radians. Orientation also is estimated using a quaternion representation, whichunlike Euler angles-is not subject to problematic singularities such as gimbal lock (Brunner, Lauffenburger, Changey, & Basset, 2015). In addition, the sensor band returns 8-channel electromyogram (EMG) data, which we used to compute descriptors of muscular effort and to estimate the movements of wrists and fingers.…”

Quaternion-based IMU and stochastic error modeling for intelligent vehicles

“…Indeed, quaternions do not suffer from the angle singularities known as the gimbal lock problem. The sensor data simulator has already been validated with real measurements considering low dynamics [7], [8]. In this paper, a validation is proposed in the intelligent vehicles context using low and high dynamics issued from real trials.…”