Experimental Evaluation of Position-Based Visual Servoing Using AprilTag for Mobile Robot Docking
摘要
Docking a mobile robot to a charging station is a critical capability for long-term autonomy, enabling continuous operation without human intervention. This paper investigates a Position-based visual servoing as a local alignment strategy for a docking a mobile robot to a charging station. In the first stage, the robot employs autonomous navigation using ROS2 NAV2 to reach an initial approach pose in front of the charging station. Once reached, a camera system detects fiducial tags placed on the station in aim to compute the relative pose. A PI controller is used to align the robot toward a precise staging pose. However, as the robot advances closer to the dock, the camera’s field of view becomes occluded, rendering visual feedback unavailable. To address this limitation, wheel odometry is integrated to maintain trajectory and ensure accurate completion of the docking maneuver. The proposed method thus leverages a hybrid sensing approach: global navigation for coarse positioning, visual tag detection for fine alignment, and odometry for blind final approach. Experimental trials demonstrate that this layered strategy achieves acceptable docking performance under realistic conditions.