Evaluating Smartphone Camera Calibration Configurations and their Effects on SLAM

Abstract. Smartphones have become increasingly viable for photogrammetric and simultaneous localization and mapping (SLAM) applications due to their portability and widespread availability. However, the repeatability of smartphone camera calibration remains a concern, as intrinsic orientation parameters (IOPs) can vary significantly between calibration attempts due to innate software and hardware corrective mechanisms. This study investigates the impact of calibration grid type, grid size, and distortion modelling on smartphone camera calibration uncertainty and its downstream effects on positioning accuracy in monocular SLAM. Using three smartphone models (iPhone 14 and two Google Pixel 7 devices), we conducted a comprehensive analysis of 24 calibration configurations processed through Kalibr. The estimated IOPs were then applied in a monocular ORB-SLAM3 pipeline, and the resulting trajectories were compared against a high-precision integrated LiDAR inertial ground truth. These findings provide insights into optimizing smartphone calibration setups, which has effects on SLAM-based applications in mobile mapping, robotics, and augmented reality.