An Improved Loopback Method for Positioning Mobile Robots in Factory Parks

The mobile robot’s LiDAR loses accuracy in the complex road system of the factory campus, and the GNSS signal is easily blocked by buildings and becomes invalid, posing serious challenges to the positioning accuracy and robustness of the mobile robot equipped with a single sensor. This research presents a multi-sensor fusion mobile robot positioning strategy based on the LIO-SAM architecture to overcome the aforementioned issues. IMU data is handled at the front end for point cloud de-distortion, and a LIDAR odometer is used to output the location. The factor graph approach is used to optimize the back end. A better loopback detection technique based on Scan-Context is suggested. It makes use of the NDT algorithm’s output to optimize the ICP algorithm’s starting value in order to rectify trajectory drift and boost the loopback detection algorithm’s effectiveness and accuracy. The strategy works well in complicated contexts, according to the study of the simulation experiment findings.