Novel Methods of Mitigating Lever Arm Effect in Redundant IMU

In this paper, two novel methods to mitigate lever arm effect in Redundant Inertial Measurement Units (RIMUs) are proposed with different approaches. With the presence of a lever arm for each sensing axis, the unexpected accelerations such as Euler and centrifugal accelerations are added to the measurements through rotational motion, resulting in an estimation error of linear acceleration. Therefore, it was previously considered as the best option to minimize the length of the lever arm and compensate the lever arm effect from the accelerometer measurements. However, this approach cannot completely remove the estimation error as the compensated value is based on the estimated angular rates, where the magnitude of the error becomes more apparent with a RIMU composed of low-grade gyroscopes that shows a higher noise level. In order to solve this problem, we propose two methods that can mitigate estimation error using the specific arrangement of the lever arm vectors and the concentrated likelihood method-based nonlinear least squares (NLS). By the proposed methods, the accuracy in compensating the lever arm effect can be increased by placing the lever arm vectors symmetrically or using the information from accelerometers when estimating angular rates. Besides, the suggested methods each have their own advantages in computational efficiency and overall navigation performance, compared to previous method. The effectiveness of the proposed methods is verified through simulations including misalignments of each redundant sensor.