MSOE: Toward Abrupt Magnetic Change in MARG Orientation Using Multi-State Estimator With Smartphones

The phone attitude angle has been used as an effective tool to many applications, which will be a tough problem when it suffers the change of external environment, such as the magnetic field abrupt variation. In this paper, we present a novel orientation estimation and calibration algorithm MSOE, a multiple states orientation estimation based on the smartphones inertial measurement units consisting of tri-axis accelerometer, gyroscope, and MARG sensors. MSOE is divided into two parts, MSOE calibration (MSOE-C) and MSOE remainder (MSOE-R). MSOE-C skillfully borrows the crowdsourcing data expressed in the form of variational quaternions measured by the phones in the same magnet field. As the non-uniform nature exists inevitably when the crowd-phones are far away from the determinand, the crowdsourcing data may be erroneous. MSOE-R provides us a reminder by warning the control center to prevent measuring by the magnetometer and mainly utilize the gyroscope for estimation during the period with unstable environment. MSOE also incorporates a compensation filter for the tri-axis sensors deviation, especially for the gyroscope drifting. Performance has been evaluated empirically in two different Android platforms with four types of phones based on the available measurement system. Performance is also benchmarked against some current estimation algorithms and MSOE provides two times improvement on the accuracy of attitude angle estimation.