Research on an extended Kalman filter-based method for autonomous localization and mapping of AGVs

Abstract

With the rapid development of industrial automation and intelligent logistics, the autonomous navigation and localization accuracy of automated guided vehicles (AGVs), as core equipment in automated transport systems, directly affects logistics efficiency and operational safety. To address insufficient AGV localization accuracy and substantial cumulative odometry errors in factory environments, this paper proposes a simultaneous localization and mapping (SLAM) method based on light detection and ranging (LiDAR) feature matching and the extended Kalman filter (EKF). First, environmental point- cloud data are acquired using LiDAR and partitioned using an adaptive distance-threshold region-segmentation algorithm. Line-segment features are then extracted as environmental landmarks by combining the Douglas–Peucker (DP) line-segment extraction algorithm with a random sample consensus (RANSAC) outlier-rejection line-fitting algorithm. Second, an interpretation-tree model is established to associate observed features with map landmarks, and a multiple-hypothesis tracking strategy is employed for initial-pose search and subsequent matching. Finally, an EKF-based filtering framework fuses LiDAR observations with odometry information to jointly estimate the AGV pose and map-landmark states. Experimental results show that the proposed method achieves a repeatability of approximately 2 cm in a workshop environment, reduces localization error by more than 85% compared with odometry-only localization, and maintains stable mapping and localization performance even in factory environments with relatively sparse features. This study provides an effective technical solution for autonomous navigation and closed-loop control of AGVs in automated transport systems.