BA-LINS: A Frame-to-Frame Bundle Adjustment for LiDAR-Inertial Navigation (2401.11491v2)

Abstract: Bundle Adjustment (BA) has been proven to improve the accuracy of the LiDAR mapping. However, the BA method has not yet been properly employed in a dead-reckoning navigation system. In this paper, we present a frame-to-frame (F2F) BA for LiDAR-inertial navigation, named BA-LINS. Based on the direct F2F point-cloud association, the same-plane points are associated among the LiDAR keyframes. Hence, the F2F plane-point BA measurement can be constructed using the same-plane points. The LiDAR BA and the inertial measurement unit (IMU)-preintegration measurements are tightly integrated under the framework of factor graph optimization. An effective adaptive covariance estimation algorithm for LiDAR BA measurements is proposed to further improve the accuracy. We conduct exhaustive real-world experiments on public and private datasets to examine the proposed BA-LINS. The results demonstrate that BA-LINS yields superior accuracy to state-of-the-art methods. Compared to the baseline system FF-LINS, the absolute translation accuracy and state-estimation efficiency of BA-LINS are improved by 29.5% and 28.7% on the private dataset, respectively. Besides, the ablation experiment results exhibit that the proposed adaptive covariance estimation algorithm can notably improve the accuracy and robustness of BA-LINS.

Introduction to BA-LINS

Bundle Adjustment (BA) has long been affirmed as a pivotal technique in enhancing the LiDAR mapping's accuracy. Traditional usages of BA, however, have largely been confined to mapping with prebuilt environments as their context rather than direct application in dead-reckoning navigation systems.

Conceptual Innovations

In this paper, BA-LINS, a frame-to-frame (F2F) BA for LiDAR-inertial navigation is proposed, exploring a novel domain that tightly integrates BA within the paradigm of LIght Detection And Ranging (LiDAR) and Inertial Measurement Unit (IMU) data. Unlike prior methodologies, BA-LINS concentrates on direct F2F point-cloud association to align the same-plane points between LiDAR keyframes, thus fostering the construction of accurate plane-point BA measurements.

System Architecture and Methodology

The system integrates LiDAR BA measurements with IMU preintegration assessments under the framework of factor graph optimization (FGO), which is a principled approach enhancing both the accuracy of navigation and robustness against perturbations in sensor readings. Additionally, BA-LINS introduces an adaptive covariance estimation algorithm explicitly designed for LiDAR BA measurements, which significantly contributes to precision.

Empirical Validation and Performance

Empirical evidence stems from exhaustive experimentation on both public and private datasets, showcasing BA-LINS's ability to outpace state-of-the-art methods in terms of absolute translation accuracy and state-estimation efficiency. Specifically, compared to the baseline system FF-LINS, BA-LINS shows a marked improvement of 29.5% in absolute translation accuracy and a 28.7% enhancement in state-estimation efficiency on private datasets. The ablation paper provides insights into the significance of the developed adaptive covariance algorithm and confirms the consistency of the system's state estimator.

Conclusion

In conclusion, the proposed frame-to-frame BA-LINS presents a substantial advancement in LiDAR-inertial navigation systems. This innovation harnesses the potential of BA beyond traditional mapping applications and demonstrates superior accuracy and computational efficiency. The integration of BA within the structure of F2F associations marks a milestone in the exploration of multi-sensor fusion navigation systems and suggests a promising avenue for future research in BA's role within navigation applications.