Abstract: The advent of the intelligent era has brought great changes to many fields. In the coal industry,the popularization
and application of autonomous truck have promoted the modernization of the industry. Simultaneous Localization and Mapping
(SLAM) technology is the key method to solve the autonomous transportation and navigation of autonomous truck. In the
underground environment,the uneven illumination conditions,the degradation of roadway characteristics,and the complex road
conditions of the working face all pose challenges to the traditional laser SLAM algorithm. Aiming at the above problems,a Li-
DAR SLAM based on multi-sensor fusion method for under roadway (ISC-LIWO) is proposed. Firstly,based on the Extended
Kalman Filter (EKF) algorithm,the wheel odometer and Inertial Measurement Unit (IMU) data are fused to achieve high-precision
odometer output,which enhances the robustness of the algorithm in underground mine environment. Secondly,the Intensity
Scan Context (ISC) is proposed as a descriptor for loop closure detection,and the candidate point cloud descriptor is
searched through a two-stage retrieval strategy based on the rotation invariance of the descriptor,which effectively improves the
retrieval speed and the accuracy of position recognition,and reduces the matching error of the downhole map. The experimental
data are obtained from the actual mining and public data sets of a coal mine in Jining City. The results reflect that the square
root of the absolute pose error of the ISC-LIWO algorithm is 33. 96 % lower than that of A-LOAM,44. 17% lower than that of
LeGO-LOAM,and 10. 02 % lower than that of LIO-SAM. Experiments show that the algorithm has higher robustness for the characteristic degradation environment of underground roadway,which can effectively reduce pose drift and map ghosting,and
provide reliable underground map and state estimation for autonomous truck.

Key words: autonomous truck,SLAM,sensor fusion,intensity information,loop closure detection