Traj-LIO: A Resilient Multi-LiDAR Multi-IMU State Estimator Through Sparse Gaussian Process

Abstract--Nowadays, sensor suits have been equipped with redundant LiDARs and IMUs to mitigate the risks associated with sensor failure. It is challenging for the previous discretetime and IMU-driven kinematic systems to incorporate multiple asynchronized sensors, which are susceptible to abnormal IMU data. To address these limitations, we introduce a multi-LiDAR multi-IMU state estimator by taking advantage of Gaussian Process (GP) that predicts a non-parametric continuous-time trajectory to capture sensors' spatial-temporal movement with limited control states. For LiDAR sensors, continuously captured points between State estimation [3] is a fundamental task in robotics, which two consecutive discrete states have to be compensated for predicts the underlying state of system through a sequence of motion distortions [45, 43] by assuming to be measured measurements from various sensors. Thus, the accuracy of LiDAR odometry heavily LiDAR has emerged as one of the most widely adopted exteroceptive depends on IMU-driven kinematics [31, 12], which are sensors for pose estimation due to its excellent range employed to not only undistort the points but also provide sensing capability in unstructured scenarios [35] and lowillumination an initial propagated state for registration. In the practical studies [15, solutions [5, 22, 20, 27, 19] have replaced the discrete-time 46, 21], the inertial-aided odometry systems [28, 30, 43, 7] states with the parametric continuous-time trajectory [13], are considered as more reliable state estimator by coupling enabling the querying of specific states given a timestamp.