Multi-Sensor Navigation System With Vision Marker Implementation
Yee Hang Wong, Ban-Hoe Kwan, Siow Cheng Chan, Oon-Ee Ng · 2025
This study presents a multi-sensor navigation system that integrates LiDAR-based SLAM with a vision-based lane tracking approach to enhance localization accuracy in indoor environments. The system is designed for Autonomous Mobile Robots (AMRs) operating in factory-like settings, where LiDAR-based localization may become unreliable due to sparse or repetitive features. The main objective is to evaluate the effectiveness of visual lane markers in improving localization consistency and reducing cumulative drift. A series of field tests were conducted in a controlled environment with four reference points, comparing localization accuracy before and after implementing lane tracking. The results show that the proposed system reduced the average Euclidean localization error from 0.272 m to 0.118 m, 56.6 % improvement. These findings demonstrate that integrating lane tracking into a multi-sensor fusion framework significantly enhances the robustness of AMR localization in challenging indoor scenarios.