Semi-Elastic LiDAR-Inertial Odometry

TL;DR

The paper tackles local inconsistency in LiDAR-inertial odometry by introducing a semi-elastic optimization that relaxes the start-state constraint between consecutive sweeps. By combining LiDAR point-to-plane constraints, IMU pre-integration, and a stable logical constraint, the method jointly optimizes beginning and end sweep states to maintain accuracy and robustness. Empirical results on four public datasets show state-of-the-art accuracy and improved consistency over traditional and elastic LIO approaches, with ablations confirming the benefits of the semi-elastic design. The approach runs in real time and is released as open-source to support community development and deployment in robotics and autonomous systems.

Abstract

Existing LiDAR-inertial state estimation assumes that the state at the beginning of current sweep is identical to the state at the end of last sweep. However, if the state at the end of last sweep is not accurate, the current state cannot satisfy the constraints from LiDAR and IMU consistently, ultimately resulting in local inconsistency of solved state (e.g., zigzag trajectory or high-frequency oscillating velocity). This paper proposes a semi-elastic optimization-based LiDAR-inertial state estimation method, which imparts sufficient elasticity to the state to allow it be optimized to the correct value. This approach can preferably ensure the accuracy, consistency, and robustness of state estimation. We incorporate the proposed LiDAR-inertial state estimation method into an optimization-based LiDAR-inertial odometry (LIO) framework. Experimental results on four public datasets demonstrate that: 1) our method outperforms existing state-of-the-art LiDAR-inertial odometry systems in terms of accuracy; 2) semi-elastic optimization-based LiDAR-inertial state estimation can better ensure consistency and robustness than traditional and elastic optimization-based LiDAR-inertial state estimation. We have released the source code of this work for the development of the community.

Figures (5)

• Figure 1: Illustration of (a) traditional optimization-based LiDAR-inertial state estimation, (b) elastic optimization-based LiDAR-inertial state estimation, (c) semi-elastic optimization-based LiDAR-inertial state estimation and (d) logical constraint.
• Figure 2: Overview of our LIO system which consists of four main modules: a pre-processing module, an initialization module, a state estimation module and a point registration module. The yellow part is the semi-elastic optimization-based LiDAR-inertial state estimation we proposed.
• Figure 3: (a) The local zigzag of trajectory estimated by traditional LiDAR-inertial state estimation. (b) The smooth trajectory estimated by our semi-elastic LiDAR-inertial state estimation.
• Figure 4: The curve of velocity magnitude estimated by three state estimation methods and measured by wheel odometer encoder on $nclt\_16$. We downsampled the sample size by 1/20 when plotting.
• Figure 5: (a) and (c) are the comparison results between our estimated trajectories and ground truth on the exemplar sequences $nclt\_7$ and $kaist\_3$. (b) and (d) are the local point cloud map of $nclt\_7$ and $kaist\_3$ respectively.