Implementation of Vision-Based Localization in Barelang-V Robot

Abstract

Reliable localization is critical needs for autonomous mobile robots in indoor and un even environment where GPS and wheel odometry are unreliable. This study implements visionbased localization system using RTAB-Map on the Barelang-V mobile robot, develop for the ABU Robot Contest. The system relies on an intel RealSense D455i RGB-D camera with built-in Inertial Measurement Unit (IMU). Visual-inertial data are processed within the ROS framework. Experimental results in an indoor environment demonstrate that the proposed approach achieves accurate and stable localization, with average positional errors of 0.051 meters when compared to ground truth measurements. These results indicate that the visual based localization with RTAB-Map pipeline provide effective solution for indoor mobile robot navigation.