Calibration and performance evaluation of low-cost IMUs

Abstract

IMUs (Inertial Measurement Units) are extensively used in many robotics applications such as navigation and mapping tasks. In almost all these systems, inertial measurements are fused with data coming from other sensors (e.g., GPS sensors, range finders, cameras, . . . ). For better results, the IMU should be carefully calibrated, in order to minimize the propagation of systematic errors. But what happens if for brief periods data coming from the other sensors are missing? Can we trust the IMU in these cases? In this paper, we present a robust and simple method to calibrate an IMU without any external equipment. We then use the calibration results to analyze the behavior of two types of MEMS based IMUs employed as a single sensor in full 3D orientation and egomotion estimation tasks.