Continuous-Time Fixed-Lag Smoothing for LiDAR-Inertial-Camera SLAM

Localization and mapping with heterogeneous multisensor fusion have been prevalent in recent years. To adequately fuse multimodal sensor measurements received at different time instants frequencies, we estimate the continuous-time trajectory by fixed-lag smoothing within a factor-graph optimization framework. With formulation, can query poses any corresponding to measurements. bound computation complexity of smoother, maintain temporal keyframe sliding windows constant size, probabilistically marginalize out control points other states, which allows preserving prior information for future sliding-window optimization. Based on smoothing, design tightly coupled SLAM algorithms variety combinations, like LiDAR-inertial LiDAR-inertial-camera systems, online offset calibration is also naturally supported. More importantly, benefiting from marginalization our derived analytical Jacobians optimization, proposed systems achieve real-time performance regardless high formulation. The widely evaluated three public datasets self-collect datasets. results demonstrate that high-accuracy pose estimations outperform existing state-of-the-art methods. benefit research community, will open source code https://github.com/APRIL-ZJU/clic .