Design, Sensing and Control of Service Robotic System for Intelligent Navigation and Operation in Internet Data Centers

Abstract

Traditionally, loading, unloading, and transporting heavy servers in Internet Data Centers (IDCs) has been a labor-intensive task, resulting in high maintenance costs and strain on workers. To address this, we design a novel, fully-automatic robot system composed of a smoothly running chassis and a strong manipulation equipment with passive compliant docking modules. This robot is capable of picking up servers (over 30 kilograms) from transfer shelves, navigating to a server cabinet, and accurately aligning itself to the cabinet to load or unload servers. After completing server maintenance operations, the robot navigates back to the transfer shelf, docks towards it, and completes the task. To achieve real-time localization, we implement a SLAM method that couples LiDAR, IMU, and wheel odometer. We also propose an efficient reflective sticker identification method to utilize artificial landmarks and thus improving localization accuracy. Finally, ArUco and QR code markers are detected to assist the robot in fine-tuning the pose for high-precision navigation tasks. Through simulation and real-world experiments, we demonstrate that our IDC robot can accomplish the necessary tasks efficiently and safely.