The method of reflection-based marker detection and identification to ensure accurate AGV docking

Area of Science:

• Robotics
• Computer Vision
• Sensor Fusion

Background:

• Automated Guided Vehicles (AGVs) require precise localization for industrial tasks like docking.
• Robust environmental perception is a significant challenge in industrial AGV applications.
• Existing localization methods may have limitations in accuracy or reliance on complex infrastructure.

Purpose of the Study:

• To develop a novel, cost-effective localization and docking method for AGVs using only 2D LiDAR.
• To enable accurate marker detection, identification, and pose estimation for precise docking.
• To validate the proposed method's performance in a heavy industrial setting.

Main Methods:

• A reflection-based marker detection and identification technique utilizing 2D Light Detection and Ranging (2D LiDAR).
• A specialized 2D marker design facilitating distance and orientation estimation.
• Integration of the marker system with AGV motion planning for autonomous docking.

Main Results:

• The proposed method accurately estimates marker distance and orientation, enabling reliable identification and docking point determination.
• Experimental validation on a heavy industrial AGV showed positional accuracy up to 1 cm and YAW orientation accuracy below 0.05 degrees.
• The AGV achieved docking precision with standard deviations below 2 cm (X, Y axes) and 1.8 degrees (YAW) at an assembly station.

Conclusions:

• The developed 2D LiDAR-based marker system offers a robust and accurate solution for AGV localization and docking in industrial environments.
• The method provides high precision, meeting the stringent requirements for automated assembly and material handling.
• This approach demonstrates the potential of 2D LiDAR for advanced AGV navigation without external infrastructure.