Design of Lidar SLAM Perception and Localization Algorithm for Underground Roadway Inspection Vehicles in Coal Mines

Abstract

The traditional laser SLAM method has rapidly increasing cumulative error, poor robustness of the rotational process, high feature association error rate, etc. in the underground environment. In order to solve the above problems, the design of LIDAR SLAM sensing and localization algorithm for coal mine underground roadway inspection vehicle is proposed. The design consists of two parts: front-end iterative Kalman filtering and back-end position map optimization. The front-end preprocesses the sensor data, constructs an observation model and a prediction model, and establishes an iterative Kalman filter, which combines the robot's a priori position through the state propagation process of prediction and observation, so as to make its a posteriori position more accurate after the state update. The back-end uses keyframes to determine whether the current position is added to the global optimization, and uses loopback detection to determine whether the current position is revisited. In addition, loopback detection and ground constraints are added to the optimization framework to optimize the relative bit positions between adjacent keyframes to ensure the consistency of the global map. The experimental results show that compared with the traditional calibration algorithm, the fusion of this algorithm with visual data has comparable accuracy, better environmental adaptation ability, and 40% higher calibration success rate.