DINS-IO: Learned Inertial Odometry via Differentiable INS Consistency

2026-07-22

AI Summary

This work addresses the reliance of learning-based inertial odometry on expensive, dense ground-truth pose annotations by proposing a self-supervised pretraining approach that requires only raw IMU measurements. The key innovation lies in formulating the velocity propagation of strapdown inertial navigation systems (INS) as a differentiable consistency prior for the first time, alongside designing a high-frequency network that directly predicts body-frame velocities. A self-supervised loss is constructed by integrating a sliding-window least-squares closed-form solver, and a low-rank adaptation (LoRA) mechanism is introduced to enable effective calibration with only a small number of labeled trajectories during fine-tuning. Experimental results demonstrate that the proposed method achieves performance on par with or even surpassing fully supervised baselines on standard benchmarks.

Abstract

The training of learned inertial odometry depends on dense, high-precision position ground truth from motion capture, visual-inertial odometry or SLAM, which is costly and hard to acquire at scale. We propose DINS-IO, which learns inertial odometry directly from raw IMU streams without position labels. Our key insight is that the strapdown INS velocity recursion is a strong, fully differentiable consistency prior: the predicted velocity, rotated into the navigation frame, must agree with the integrated specific force up to an unknown initial velocity and a constant accelerometer bias. We cast this constraint as a sliding-window least-squares problem with a globally shared bias, solve it in closed form, and use the solver residual as a self-supervised loss whose gradient flows back to the network through the analytic solution. To supply this per-sample constraint, we design a high-frequency network that emits dense body-frame velocity at the IMU rate. Since the self-supervised network learns consistent motion but its velocity is not yet metrically calibrated, we calibrate it to true metric velocity from a few labeled trajectories by directly supervising the predicted body-frame velocity and adapting only low-rank (LoRA) patches. On standard benchmarks, DINS-IO pretrained self-supervised and fine-tuned with a small fraction of labels matches or surpasses fully supervised baselines.

Problem

• inertial odometry
• position ground truth
• self-supervised learning
• IMU
• metric calibration

Innovation

• Differentiable INS
• Self-supervised Learning
• Inertial Odometry
• Closed-form Solver
• LoRA Calibration