arXiv cs.CL· Yaru Gu, Ting Zou·· 10 小时前AI 评分27
DFIO：将误差状态卡尔曼滤波更新步骤嵌入计算图的可微分滤波惯性里程计框架
Differentiable Filtering for Inertial Odometry
AI 导读
DFIO 将误差状态卡尔曼滤波的更新阶段直接嵌入计算图，在独立预训练后通过可微分测量更新对位移与协方差输出头做联合优化。在行人运动数据集上，其平均绝对平移误差和 1 秒相对平移误差较 TLIO 基线分别降低 13.2% 和 39.6%，较自身预训练版本分别降低 7.7% 和 17.4%。在未校准智能手机惯性数据上，该方法也能有效抑制平移漂移并保持轨迹对齐，源代码将公开。
正文 · 原文
Abstract:Accurate inertial odometry is critical for autonomous systems but suffers from unbounded drift caused by the integration of sensor noise and biases. In approaches such as TLIO, neural pseudo-measurements are trained independently of their downstream filter. This paper introduces Differentiable Filtering for Inertial Odometry (DFIO), a learning framework that embeds the update stages of an error-state Kalman filter directly into a computational graph. Following standalone pre-training, DFIO jointly optimizes the displacement and covariance output heads using posterior-position supervision through differentiable measurement updates. The proposed system is validated on public pedestrian motion datasets featuring both calibrated and uncalibrated inertial measurements. Experimental results on a pedestrian dataset demonstrate that DFIO reduces mean absolute and one-second relative translation errors by 13.2% and 39.6% relative to the TLIO baseline, and by 7.7% and 17.4% relative to its pre-trained counterpart. Furthermore, evaluations on uncalibrated smartphone inertial data indicate that the proposed method effectively mitigates translational drift and maintains trajectory alignment under highly noisy sensor conditions. Source code will be made publicly available.
| Comments: | 19 pages, 4 figures |
| Subjects: | Robotics (cs.RO); Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.13517 [cs.RO] |
| (or arXiv:2611.13517v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2611.13517 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yaru Gu [view email]
[v1]
Tue, 10 Nov 2026 03:45:28 UTC (1,016 KB)
来源：arXiv cs.CL · arxiv.org