SLICT: Multi-Input Multi-Scale Surfel-Based Lidar-Inertial Continuous-Time Odometry and Mapping

While feature association to a global map has significant benefits, keep the computations from growing exponentially, most lidar-based odometry and mapping methods opt associate features with xmlns:xlink="http://www.w3.org/1999/xlink">local maps at one voxel scale. Taking advantage of fact that xmlns:xlink="http://www.w3.org/1999/xlink">surfels (surface elements) different scales can be organized in tree-like structure, we propose an octree-based global map xmlns:xlink="http://www.w3.org/1999/xlink">multi-scale surfels updated incrementally. This alleviates need for recalculating, example, k-d tree whole repeatedly. The system also take input single or number sensors, reinforcing robustness degenerate cases. We point-to-surfel (PTS) scheme, continuous-time optimization on PTS IMU preintegration factors, along loop closure bundle adjustment, making complete framework Lidar-Inertial mapping. Experiments public in-house datasets demonstrate advantages our compared other state-of-the-art methods.