A Lightweight LiDAR SLAM in Indoor-Outdoor Switch Environments

Simultaneous Localization and Mapping (SLAM) can provide pose estimation and map information. It is widely used in Intelligent transportation systems such as smart vending vehicles. However, existing SLAM methods used for vending vehicles rarely focus on indoor environments. We proposed a real-time LiDAR-based SLAM with high accuracy in both outdoor and indoor environments. Our method takes the advantage of Inertial Measurement Unit (IMU) to reduce the distortion of raw data. Corner and planar features are extracted for point cloud registration. Besides, different optimization formulas are applied in different scenes. The proposed method achieves an average error of fewer than 1m in the KITTI Odometry benchmark and has high accuracy in different experiments.