Multi-sensor Calibration System for Mobile Robots
Wei Wu,Chang Liu,Wenhan Cao,Chao Chen,Hongbo Li,Shengbo Eben Li
DOI: https://doi.org/10.1109/cvci59596.2023.10397235
2023-01-01
Abstract:Multi-sensor calibration is crucial for mobile robots, especially in logistics and warehouse environments. While calibration techniques exist for individual sensors, industry-level calibration of the multiple sensors on mobile robots covering the entire operational life cycle is rarely explored. To bridge the gap, we present a fast, accurate, and scalable calibration system supporting both manufacturing and operation phases for mobile robots with diverse sensors like odometers, IMUs, LiDARs, and cameras. Specifically, we propose an online yaw angle estimation approach for IMU by fusing odometer data as compensation. Additionally, we pioneer an approach to calibrate 2D LiDAR using surrounding objects’ reflectance strength as the calibration metric in the manufacturing phase calibration, and the iterative closest point method for refinements during operation phase calibration. Furthermore, we develop a least squares formulation to calibrate cameras’ intrinsic and extrinsic parameters, achieving rapid convergence. The system has been used by Geekplus across factories worldwide including Asia, Europe, America, and Oceania, successfully calibrating and deploying over 30,000 mobile robots.