Wheel-Inertial Fusion Localization Strategy for Mobile Robots in Confined Pipeline Spaces

Pipeline robots hold significant promise in fields like large-scale industrial manufacturing and electric energy production. However, their current limitations in precise positioning and autonomous operation are primarily due to the constraints of the narrow and confined pipeline environments. To address the common issue of noise interference with conventional inertial positioning systems, we propose an enhanced wheel-inertial fusion positioning framework aimed at achieving effective robot positioning for short to medium-range travel. This method comprises a front-end positioning module and a parallel-running back-end positioning module. In the front-end positioning, robot pose is calculated based on data from inertial measurement units (IMU) and wheel encoders, utilizing the error state Kalman filter (ESKF) fusion algorithm and graph optimization. The back-end module focuses on global optimization of historical poses and inertial parameters through graph optimization. Additionally, the system estimates certain inertial parameters during the system's initialization phase. Experimental results show that the method presented in this paper effectively improves the robot's positioning accuracy in closed pipelines.