NIF-VIO: Visual--Inertial Odometry Across Inertial and Non-Inertial Reference Frames
Organizations: Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai 200083, China · University of Chinese Academy of Sciences, Beijing 100049, China · GravityXR, Ningbo 315040, China · Advanced Display and Sensing Research Center, Yongjiang Laboratory, Ningbo 315201, China ·
Abstract
Conventional visual--inertial odometry (VIO) assumes that observations from cameras and an inertial measurement unit (IMU) are jointly consistent in an inertial reference frame. However, this assumption breaks down under non-inertial conditions. Inside accelerating or rotating carriers, visual observations of the interior reflect carrier-relative motion, while IMU measurements contain additional carrier-induced components, which can compromise conventional fusion. We therefore present NIF-VIO for localization relative to the local environment across inertial and non-inertial reference frames. It adapts visual--inertial fusion based on measurement consistency. As inertial conditions change, early constraint isolation is combined with local reoptimization to limit error propagation before inconsistency confirmation and repair the affected local geometry. Under non-inertial conditions, a compact model with temporally shared parameters is introduced to account for carrier motion while preserving inertial information that complements visual constraints. Experiments on public benchmarks and a self-collected dataset demonstrate that NIF-VIO achieves competitive performance under inertial conditions and improved localization accuracy and availability under non-inertial conditions.
Figures & tables
| Method | Machine Hall | Vicon Room | TUM-VI Room | |||
|---|---|---|---|---|---|---|
| ATE | ARE | ATE | ARE | ATE | ARE | |
| (m) | ( ∘ ) | (m) | ( ∘ ) | (m) | ( ∘ ) | |
| VINS-Fusion | 0.282 | 1.945 | 0.148 | 3.232 | 0.071 | 1.171 |
| OpenVINS | 0.138 | 1.674 | 0.065 | 1.656 | 0.074 | 1.984 |
| Basalt | 0.088 | 1.074 | 0.076 | 1.317 | 0.066 | 1.989 |
| OKVIS2 | 0.087 | 1.261 | 0.043 | 1.403 | 0.044 | 1.139 |
| Mode | Method | Car (5 seq.) | Subway (5 seq.) | Elevator (5 seq.) | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| C-ATE (m) | C-ARE ( ∘ ) | CR (%) | C-ATE (m) | C-ARE ( ∘ ) | CR (%) | C-ATE (m) | C-ARE ( ∘ ) | CR (%) | ||
| VO | VINS-Fusion | 0.038 | 3.413 | 64.83 | 0.081 | 2.627 | 83.09 | 0.125 | 2.786 | 77.85 |
| Basalt | 0.042 | 2.198 | 57.45 | 0.085 | 2.072 | 56.46 | 0.142 | 4.174 | 70.21 | |
| OKVIS2 | 0.020 | 0.861 | 68.95 | 0.031 | 0.670 | 72.31 | 0.089 | 1.685 | 84.78 | |
| ORB-SLAM3 | 0.019 | 1.208 | 93.43 | 0.034 | 1.008 | 94.83 | 0.094 | 1.495 | 89.42 | |
| VIO | VINS-Fusion | 0.037 | 1.279 | 38.72 | 0.131 | 1.478 | 20.76 | 0.111 | 1.944 | 35.54 |
Visual-Inertial Odometry
Visual-inertial odometry estimates camera motion by combining image geometry with inertial measurements for metric navigation.
See all 64 papers