Author

Chunxi Xia

View on Semantic Scholar

2 papers indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

2026

GREAT Dataset: A Multi-Sensor Raw Observation Dataset for High-Precision Urban Navigation

In recent years, multi-sensor fusion technology has emerged as a key approach for high-accuracy localization in intelligent driving. Although numerous multi-sensor datasets have been published, few of them pay enough attention to providing high-precision data, which is essential for assisting integrated methods to achieve the stringent demands of localization. To address this gap, we present the GREAT Dataset, a multi-sensor dataset with raw observations collected on a vehicle-mounted platform in urban environments. The sensor suite comprises a multi-frequency, multi-constellation Global Navigation Satellite System (GNSS) receiver, a Micro-Electro-Mechanical System (MEMS) Inertial Measurement Unit (IMU), a tactical-grade IMU, a stereo camera, and a 16-beam Light Detection and Ranging (LiDAR). All of these sensors are hardware synchronized to the Global Positioning System (GPS) time. The proposed dataset provides both pseudorange and carrier-phase observations of GNSS, along with the tactical-grade IMU measurements. Eight sequences are contained in the GREAT Dataset, captured in urban environments like campuses, urban canyons, and suburban areas. While such diverse sequences pose challenges to different types of sensors, the high-quality data remains potential for algorithms to achieve high-accuracy positioning results, thereby meeting the localization requirements of intelligent driving applications. Besides, we conducted evaluations on a series of representative algorithms to confirm the validity of the dataset. The results also demonstrate that our dataset is suitable for testing multi-sensor integrated approaches. To benefit the research community, we have made the data and relevant scripts publicly available. The download links could be found at https://github.com/GREAT-WHU/GREAT-Dataset Note to Practitioners—This work aims to support multi-sensor fusion algorithms in achieving the strict requirements of localization in autonomous driving. The existing open-sourced multi-sensor datasets often lack sufficient consideration of data precision, which is crucial for accurately evaluating positioning capabilities. To bridge this deficiency, we propose the GREAT dataset, a high-precision multi-sensor dataset with raw observations collected in urban environments. By utilizing this dataset, evaluations can be conducted through comparison of positioning results against our provided high-precision reference solutions. Furthermore, to facilitate researchers’ efforts, we have performed evaluations on a series of representative algorithms. The results demonstrate that the GREAT dataset not only serves as a robust benchmark for assessing localization performance but also presents challenges for achieving high-accuracy positioning, thereby promoting advances in integrated algorithm development.

Xingxing Li, Siqi Chen, Chunxi Xia et al. · IEEE Transactions on Automat... · 0 citations

2026

An Equivariant Filter-Based State Estimator for Tightly-Coupled LiDAR–Radar–Inertial Odometry

Over the past decade, the fusion of light detection and ranging (LiDAR) and inertial navigation systems (INS) has become a reliable solution for environmental perception in intelligent mobile platforms. However, LiDAR performance degrades under adverse weather conditions. Millimeter-wave radar offers complementary benefits: it is robust to environmental disturbances and provides direct Doppler velocity measurements. Moreover, most existing fusion frameworks rely on the extended Kalman filter (EKF), which suffers from linearization errors and inconsistency. To address these limitations, we propose and derive a tightly-coupled LiDAR–radar–inertial odometry framework based on the equivariant filter (EqF). The framework leverages both LiDAR and radar to provide complementary constraints on the 9-dimensional navigation state, significantly improving accuracy and robustness in challenging environments. Furthermore, we conduct an in-depth observability analysis using Lie derivative theory, examining both the nonlinear system and its discrete filter system. We show that the EqF preserves the same unobservable directions as the underlying nonlinear system, thereby ensuring consistent state estimation, a property that standard EKF does not possess. In addition, experiments on real-world datasets demonstrate that our method outperforms the state-of-the-art LiDAR–inertial odometry approach and other EKF-based approaches in terms of localization accuracy, robustness and velocity estimation precision. Note to Practitioners—This work aims to provide a high-precision and robust fusion framework for LiDAR, radar, and INS in intelligent mobile platforms such as autonomous vehicles and drones. On such platforms, most existing EKF-based LiDAR-inertial odometry methods suffer from performance degradation under adverse weather conditions and from the inherent inconsistency of the EKF. To address these challenges, we propose a tightly-coupled LiDAR–radar–inertial odometry framework based on the EqF. The approach exploits the symmetry of the semi-direct product group to jointly model the navigation state and IMU biases within a geometrically consistent structure. We further conduct an in-depth observability analysis, demonstrating that the proposed framework preserves the correct unobservable directions of the underlying nonlinear system. As a result, the system exhibits superior robustness against initialization errors and external disturbances, along with improved accuracy compared to traditional approaches.

Anbo Tao, Yarong Luo, Chi Guo et al. · IEEE Transactions on Automat... · 0 citations