Fused Multi-IMU Direct Sensor-to-Vehicle Extrinsic Calibration

Gregory Mifflin, David Bevly

Abstract:	Abstract—Extrinsic sensor calibration is a crucial step in obtaining useful navigation information for an autonomous vehicle. However, manual calibration is costly and must be performed by a trained professional. An online autonomous calibration routine would resolve these issues, and allow for dynamic reconfiguration of sensors by the end user. The primary barrier to such a calibration routine is the lack of a reliable Sensor-to-Vehicle calibration method around which the other Sensor-to-Sensor calibrations can be referenced. Previous work has focused on generating a feasible method of estimating IMU-to-Vehicle extrinsic calibration parameters with limited information. However, none of these methods have yet achieved accuracy at or above the industry standard. This work proposes a novel method to improve direct IMU-to-Vehicle extrinsic calibration by fusing information from multiple IMUs on board the vehicle. A weighted nonlinear least squares approach is used to simultaneously estimate the calibration parameters of a number of IMUs relative to the vehicle frame. Each pair of IMUs generates an additional IMU-to-IMU calibration that is weighted according to the error characteristics of the measurements in the least squares estimator. These IMUto-IMU calibrations constrain the combined IMU-to-Vehicle pose, resulting in improved IMU-to-Vehicle pose estimates. A battery of tests in simulation was conducted to validate the proposed Multi-IMU calibration method, and assess its performance as a function of the number of IMUs mounted on the vehicle, as well as the error characteristics of those IMUs. Index Terms—IMU, extrinsic calibration, pose estimation
Published in:	2023 IEEE/ION Position, Location and Navigation Symposium (PLANS) April 24 - 27, 2023 Hyatt Regency Hotel Monterey, CA
Pages:	673 - 679
Cite this article:	Mifflin, Gregory, Bevly, David, "Fused Multi-IMU Direct Sensor-to-Vehicle Extrinsic Calibration," 2023 IEEE/ION Position, Location and Navigation Symposium (PLANS), Monterey, CA, April 2023, pp. 673-679. https://doi.org/10.1109/PLANS53410.2023.10140061
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In