UNRIO: Uncertainty-Aware Velocity Learning for Radar-Inertial Odometry

TL;DR

UNRIO is an innovative radar-inertial odometry system that uses a transformer neural network to directly estimate ego-velocity from raw radar signals, incorporating uncertainty estimation for improved pose accuracy.

Contribution

It introduces a transformer-based neural network that processes raw radar spectral data for velocity estimation and integrates uncertainty calibration into odometry.

Findings

Achieves lowest relative pose error on most sequences in the IQ1M dataset.

Outperforms classical DSP baselines, especially in lateral-motion scenarios.

Successfully propagates uncertainty estimates into pose graph optimization.

Abstract

We present UNRIO, an uncertainty-aware radar-inertial odometry system that estimates ego-velocity directly from raw mmWave radar IQ signals rather than processed point clouds. Existing radar-inertial odometry methods rely on handcrafted signal processing pipelines that discard latent information in the raw spectrum and require careful parameter tuning. To address this, we propose a transformer-based neural network built on the GRT architecture that processes the full 4-D spectral cube to predict body-frame velocity in two modes: a direct linear velocity estimate and a per-anglebin Doppler velocity map. The network is trained in three stages: geometric pretraining on LiDAR-projected depth, velocity or Doppler fine-tuning, and uncertainty calibration via negative log-likelihood loss, enabling it to produce uncertainty estimates alongside its predictions. These uncertainty estimates are propagated into a sliding-window pose graph that fuses radar velocity factors with IMU preintegration measurements. We train and evaluate UNRIO on the IQ1M dataset across diverse indoor environments with both forward and lateral motion patterns unseen during training. Our method achieves the lowest relative pose error on the majority of sequences, with particularly strong gains over classical DSP baselines on Lateral-motion trajectories where sparse point clouds degrade conventional velocity estimators.