UL-VIO: Ultra-lightweight Visual-Inertial Odometry with Noise Robust Test-time Adaptation

TL;DR

UL-VIO introduces an ultra-lightweight visual-inertial odometry network under 1 million parameters, capable of effective test-time adaptation for robustness against environmental shifts, suitable for resource-constrained devices.

Contribution

The paper presents a highly compressed VIO network with a novel test-time adaptation method using BatchNorm updates based on pseudo labels, addressing environmental variability and resource limits.

Findings

36X smaller network size than state-of-the-art with minimal accuracy loss

Effective noise-robust test-time adaptation demonstrated on multiple datasets

First to perform noise-robust TTA on VIO systems

Abstract

Data-driven visual-inertial odometry (VIO) has received highlights for its performance since VIOs are a crucial compartment in autonomous robots. However, their deployment on resource-constrained devices is non-trivial since large network parameters should be accommodated in the device memory. Furthermore, these networks may risk failure post-deployment due to environmental distribution shifts at test time. In light of this, we propose UL-VIO -- an ultra-lightweight (<1M) VIO network capable of test-time adaptation (TTA) based on visual-inertial consistency. Specifically, we perform model compression to the network while preserving the low-level encoder part, including all BatchNorm parameters for resource-efficient test-time adaptation. It achieves 36X smaller network size than state-of-the-art with a minute increase in error -- 1% on the KITTI dataset. For test-time adaptation, we propose to use the inertia-referred network outputs as pseudo labels and update the BatchNorm parameter for lightweight yet effective adaptation. To the best of our knowledge, this is the first work to perform noise-robust TTA on VIO. Experimental results on the KITTI, EuRoC, and Marulan datasets demonstrate the effectiveness of our resource-efficient adaptation method under diverse TTA scenarios with dynamic domain shifts.