LiDAR-Inertial Odometry Based on Extended Kalman Filter

TL;DR

This study reframes LiDAR-Inertial Odometry as recursive Bayesian estimation solved by an Extended Kalman Filter. It uses IMU preintegration for prediction and LiDAR-local-map scan matching as the observation model, enabling a loosely-coupled LIO (KLIO) that updates pose, velocity, and IMU biases through EKF. The approach achieves precise trajectory and map estimates on the Newer College dataset, with competitive accuracy against state-of-the-art tightly- and loosely-coupled methods, and demonstrates robustness to aggressive rotations and long loops. The work highlights the viability of EKF-based LIO architectures and outlines avenues for reliability estimation in LIO systems.

Abstract

LiDAR-Inertial Odometry (LIO) is typically implemented using an optimization-based approach, with the factor graph often being employed due to its capability to seamlessly integrate residuals from both LiDAR and IMU measurements. Conversely, a recent study has demonstrated that accurate LIO can also be achieved using a loosely-coupled method. Inspired by this advancements, we present a LIO method that leverages the recursive Bayes filter, solved via the Extended Kalman Filter (EKF) - herein referred to as KLIO. Within KLIO, prior and likelihood distributions are computed using IMU preintegration and scan matching between LiDAR and local map point clouds, and the pose, velocity, and IMU biases are updated through the EKF process. Through experiments with the Newer College dataset, we demonstrate that KLIO achieves precise trajectory tracking and mapping. Its accuracy is comparable to that of the state-of-the-art methods in both tightly- and loosely-coupled methods.

Figures (5)

• Figure 1: A mapping result using KLIO on the 02_long_experiment from the Stereo-Cam dataset, part of the Newer College dataset ramezani2020newer, demonstrates KLIO's effectiveness. Despite the absence of a loop-closure process, it successfully closes large loops, with the total walking distance exceeding 3 km. The estimated trajectory is illustrated by the yellow line, and the colors of the points reflect the LiDAR intensity. The bottom left detailed view reveals the precise reconstruction of cars, trees, and windows.
• Figure 2: System overview of LiDAR-Inertial Odometry based on Extended Kalman Filter (KLIO).
• Figure 3: The evaluation of absolute pose error (APE) using evo grupp2017evo against the estimates made by KLIO on the 02_long_experiment sequence.
• Figure 4: The local map construction on the 01_short_experiment sequence with $\gamma_{\rm th}$ set to 0.5, resulting in the registration of only 15 keyframes. The points color indicates height from the origin.
• Figure 5: Mapping results using KLIO on data that involved aggressive spinning, where the maximum angular velocities measured by the IMU exceeded $4~{\rm rad/sec}$. The colors represent intensity levels, with blue and red indicating low and high values.