​​​ AirIMU: Learning Uncertainty Propagation for Inertial Odometry

AirIMU: Learning Uncertainty Propagation for Inertial Odometry

Yuheng Qiu1, Chen Wang2, Can Xu1, Yutian Chen1,
Xunfei Zhou3, Youjie Xia3, Sebastian Scherer1
1Robotics Institute, Carnegie Mellon University, 2Spatial AI & Robotics Lab, University at Buffalo, 3OPPO US Research Center
arXiv GitHub Video

Abstract

Accurate uncertainty estimation for inertial odometry is the foundation to achieve optimal fusion in Inertial odometry (IO) using strap-down inertial measurement units (IMUs) is critical in many robotic applications where precise orientation and position tracking are essential. Prior kinematic motion model-based IO methods often use a simplified linearized IMU noise model and thus usually encounter difficulties in modeling non-deterministic errors arising from environmental disturbances and mechanical defects. In contrast, data-driven IO methods struggle to accurately model the sensor motions, often leading to generalizability and interoperability issues. To address these challenges, we present AirIMU, a hybrid approach to estimate the uncertainty, especially the non-deterministic errors, by data-driven methods and increase the generalization abilities using model-based methods. We demonstrate the adaptability of AirIMU using a full spectrum of IMUs, from low-cost automotive grades to high-end navigation grades. We also validate its effectiveness on various platforms, including hand-held devices, vehicles, and a helicopter that covers a trajectory of 262 kilometers. In the ablation study, we validate the effectiveness of our learned uncertainty in an IMU-GPS pose graph optimization experiment, achieving a 31.6\% improvement in accuracy. Experiments demonstrate that jointly training the IMU noise correction and uncertainty estimation synergistically benefits both tasks.

Method

Left: Traditional model-based method. Middle: Our proposed AirIMU. Right: Learning-based algorithms.

Experiments

We present the comparison across five datasets' sensors in terms of IMU's price (Unit: \$) and navigation performance categories. Navigation performance is evaluated based on the gyroscope’s in-run bias stability metric (Unit:$^{\circ}$/hr).

IMU-Centric Graph Optimization on EuRoC

Ablation study of the IMU-centric GPS PGO. In this experiment, the GPS frequency is set to 0.1 Hz.

Subt-MRS

During the stationary period (highlighted with a black box), applying the Average Bias correction actually degrades the integration performance compared to using raw data, while the AirIMU successfully captures the offset of the IMU during the stationary period.

KITTI

Left: The ROE (Unit: $\deg$) and RTE (Unit: $\mathrm{m}$) of IMU integration over 1 second (10 frames) on test dataset. RoNIN refers to the RoNIN ResNet-50 model specifically. Right: Trajectory 0022 in KITTI-Raw dataset estimated by RoNIN, Baseline and AirIMU.

BibTeX


      @article{qiu2023airimu,
            title={AirIMU: Learning Uncertainty Propagation for Inertial Odometry}, 
            author={Yuheng Qiu and Chen Wang and Xunfei Zhou and Youjie Xia and Sebastian Scherer},
            year={2023},
            eprint={2310.04874},
            archivePrefix={arXiv},
            primaryClass={cs.RO}
      }