Lightweight Generator of Synthetic IMU Sensor Data for Accurate AHRS Analysis

Accurate orientation estimation is crucial in many application areas, including unmanned ground and aerial navigation for industrial automation and human motion tracking for human-robot interaction. State-of-the-art techniques leverage Inertial Measurement Units (IMU) due to their small size, low energy footprint, and ever-increasing accuracy, which provide Magnetic, Angular Rate, and Gravity (MARG) sensor measurements. Available attitude determination techniques rely on advanced signal processing algorithms to compensate for the gyroscope integration drift. The comparison of different algorithms depends solely on the collected ground-truth data set, which is difficult to replicate. This paper introduces a lightweight software framework to generate synthetic IMU sensor data. We generate the ground-truth orientation of the sensor body frame and apply an inverse navigation process to obtain corresponding synthetic sensor data. Additionally, we compare two well-known orientation estimation algorithms applied to the synthetically generated data from our framework. Evaluation results demonstrate that the proposed software framework represents a fast and easy-to-use solution to the problem of evaluation of different orientation estimation algorithms while providing access to ground truth measurements.