Targetless intrinsic calibration for rotating LiDAR systems: a physical decoupling protocol for high-precision measurement in geometrically constrained environments

Abstract

Digital-twin construction for underground space relies on high-fidelity 3D geometric data. In GNSS-denied environments such as tunnels, economical mobile-scanning systems exhibit severe geometric distortions and trajectory drifts caused by sensor-intrinsic parameter coupling. This study develops a targetless self-calibration framework by leveraging physical-experiment decoupling. By investigating the rank-deficiency of the intrinsic Jacobian-matrix, this protocol restores parameter observability and eliminates systematic geometric distortions. Validation in scenarios simulating tunnel features confirms that the lightweight hardware configuration accurately estimates 3D lever-arm offsets. The point-cloud consistency Root Mean Square Error reaches 3.6 mm. Furthermore, systematic drift is mitigated by 87.2%. This modeling significantly improves the precision of low-cost sensing nodes for the digital perception of underground facilities.