Towards Fast Correspondence-Free Odometry Using Multiple FMCW Lidars
- 23 July 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (9) , 9088-9095
Abstract
3D FMCW lidars return relative velocity measurements via the Doppler effect, which provides a new form of information for motion estimation. In our prior work, we proposed an odometry method that avoids the conventional ICP-based approach and uses the Doppler velocity measurements in a correspondence-free way. The caveat is that the angular velocity is not observable from the Doppler measurements of a single lidar, for which we compensate by using a gyroscope. This paper is an extension of our prior work that furthers our experiments using multiple FMCW lidars. We first present better odometry performance using an improved calibration for the Doppler measurement bias and noise, and additionally show how the Expectation-Maximization algorithm can be applied to calibrate without the groundtruth trajectory. We then show improved results on our correspondence-free odometry when given access to multiple lidars and gyroscopes. We also present preliminary results without gyroscope sensors to validate our prior algebraic study that the 6-degrees-of-freedom motion is still observable with multiple lidars. While we show that odometry performance without gyroscopes is not yet as accurate, we compare against simulated measurements to motivate the potential of Doppler-only lidar odometry as a future research direction.Keywords
This publication has 44 references indexed in Scilit:
- State Estimation for Robotics2024
- LION: Lidar-Inertial Observability-Aware Navigator for Vision-Denied Environments2021
- Learning a Bias Correction for Lidar-Only Motion Estimation2018
- Gaussian process estimation of odometry errors for localization and mapping2017
- PROBE-GK: Predictive robust estimation using generalized kernels2016
- Low-drift and real-time lidar odometry and mappingAutonomous Robots, 2016
- Instantaneous ego-motion estimation using multiple Doppler radars2014
- CELLO: A fast algorithm for Covariance Estimation2013
- Are we ready for autonomous driving? The KITTI vision benchmark suite2012
- A method for registration of 3-D shapesIEEE Transactions on Pattern Analysis and Machine Intelligence, 1992