Adaptive Kalman Filter-based Fusion of Multiple Low-cost IMUs for Mobile Robot Navigation

초록

Inertial measurement unit (IMU) provides inertial data at high rates without external signals, and with the advancement of MEMS technology, they are widely used for estimating the position and attitude of mobile robots. However, low-cost MEMS IMUs are susceptible to errors and noise. As an alternative to this limitation, a fusion approach based on a virtual IMU (VIMU), which combines measurements from multiple IMUs, is proposed. Prior studies that construct VIMUs mostly use averaging or a standard Kalman filter (KF) with fixed noise covariances, which leads to performance degradation under fast dynamic maneuvers. To address this, we propose the multiple low-cost IMU fusion method using an adaptive Kalman filter (AKF) that can adjust the process noise. The position estimation accuracy of the proposed method was evaluated through extensive simulation studies and experimental tests using a mobile robot platform. Both simulation and experimental results show that the proposed method showed a clear advantage under fast dynamic maneuver while also maintaining a reasonable level of accuracy in slow dynamic maneuver. These results demonstrate the potential of the proposed fusion algorithm as a practical solution for navigation in environments with frequent maneuver changes.

키워드