A Single 2-D LiDAR Extrinsic Calibration for Autonomous Mobile Robots

• Published in: IEEE transactions on instrumentation and measurement 2023, Vol.72, p.1-9
• Main Authors: Van Toan, Nguyen, Khoi, Phan Bui, Yi, Soo-Yeong
• Format: Article
• Language: English
• Subjects: Autonomous mobile robots (AMRs); Calibration; Cameras; Collision avoidance; extrinsic calibration; Geometric constraints; iterative least square; Laser radar; Lidar; Manufacturing; Mobile robots; Multisensor fusion; Object recognition; Parameters; Position sensing; Robot kinematics; robot operating system (ROS)-based coordinate transformation; Robot vision systems; Robots; Sensors; single 2-D light detection and ranging (LiDAR)
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2023.3325856

Autonomous mobile robots (AMRs) have revolutionized various aspects of our daily lives and manufacturing services. To enhance their efficiency, productivity, and safety, AMRs are equipped with advanced capacities such as object detection and tracking, localization, collision-free navigation, and decision-making. Among these technologies, 2-D light detection and ranging (LiDAR) commonly stands out as the prevailing choice, showcasing remarkable accomplishments in practice. Obviously, the precision of the mentioned modules is affected by the accuracy of 2-D LiDAR observed data. Typically, 2-D LiDAR intrinsic parameters are adequately calibrated during the manufacturing process, while the extrinsic parameters should be intervened by the user at the application level. Previous research has predominantly emphasized extrinsic calibration for sensor fusion, given its perceived appeal over individual 2-D LiDAR extrinsic calibration. However, it is important to note that a multisensor system usually includes more favorable geometric constraints between different sensor datasets. In contrast, a 2-D LiDAR sensor only provides position information in a 2-D horizontal plane, resulting in fewer features or constraints when used alone. Besides, in the realm of multisensor calibration, the direct incorporation of observed data within the robot base coordinates is often overlooked, despite it is necessary for AMR applications. This article presents an extrinsic calibration for coordinates of a single 2-D LiDAR in AMRs' base coordinates directly, which ensures accuracy as well as easy tool installation, fast, and simple observation for data samples without supports from other sensors. The proposed method has been verified through both simulation and real experiments.