AirIO: Learning Inertial Odometry with Enhanced IMU Feature Observability
Yuheng Qiu, CAN XU, Yutian Chen, Shibo Zhao, Junyi Geng, Sebastian Scherer
AI summary
Problem
Learning-based inertial odometry methods fail to generalize to highly dynamic UAV flights because they transform raw IMU data to global coordinates, which obscures critical attitude and motion observability.
Approach
AirIO retains raw IMU measurements in the body frame, explicitly encodes drone attitude, and fuses a learned motion network with an uncertainty-aware Extended Kalman Filter to predict velocity and estimate pose.
Key results
- 66.7% average accuracy gain from body-frame IMU representation
- 23.8% additional improvement by explicitly encoding attitude
- Outperforms state-of-the-art methods without external sensors or control inputs
- Strong generalizability to unseen datasets and robustness under aggressive maneuvers
Why it matters
Enables lightweight, cost-effective UAV navigation using only IMU data, making accurate agile flight viable in resource-constrained real-world applications.
Abstract
Inertial odometry (IO) using only Inertial Mea- surement Units (IMUs) offers a lightweight and cost-effective solution for Unmanned Aerial Vehicle (UAV) applications, yet existing learning-based IO models often fail to generalize to UAVs due to the highly dynamic and non-linear-flight patterns that differ from pedestrian motion. In this work, we identify that the conventional practice of transforming raw IMU data to global coordinates undermines the observability of critical information in UAVs. By preserving the body-frame representation, our method achieves substantial performance improvements, with a 66.7% average increase in accuracy across three datasets. Furthermore, explicitly encoding attitude information into the motion network results in an additional 23.8% improvement over prior results. Combined with a data-driven IMU correction model (AirIMU) and an uncertainty-aware Extended Kalman Filter (EKF), our approach ensures robust state estimation under aggressive UAV maneuvers without relying on external sensors or control inputs. Notably, our method also demonstrates strong generalizability to unseen data, underscoring its potential for real-world UAV applications.