Video of the manuscript "A Hybrid-Dimensional Laser SLAM Framework for Indoor Quadruped Inspection Robots"

Due to the rapid mobility and superior obstacle surmounting capabilities, quadruped robots are increasingly employed in industrial inspections. Quadruped robots usually utilize laser simultaneous localization and mapping (SLAM) for autonomous navigation. However, SLAM is susceptible to inaccuracies under the rapid movement and rotation of quadruped robots. To address this challenge, this paper proposes a hybrid-dimensional laser SLAM framework for indoor quadruped robots, which operates solely using 3D hybrid solid-state LiDAR and inertial measurement unit (IMU), thereby obviating the necessity for supplementary sensors. The framework facilitates high-bandwidth and high-frequency dead-reckoning using laser inertial odometry (LIO) based on point-to-map matching. This prior for scan matching, along with motion blur removed, multi-frame fused and 2D compressed LiDAR scans, are input into graph-based SLAM to obtain optimized 2D grid map and pose estimation. Optimized poses further enhance the accuracy of the 3D point cloud map generated by LIO. The effectiveness of the framework is demonstrated through experiments conducted with the quadruped robot in a simulated train depot inspection scenario. Compared to graph-based SLAM without and with legged odometry as scan matching prior, the proposed framework exhibits enhanced mapping accuracy and localization precision.