Spatial locating of magnetic mobile robot in closed steel box girder based on UWB sensors and structural boundary

Abstract Fatigue cracks in orthotropic steel decks pose a significant threat to bridge safety, yet traditional inspection methods are inefficient and inaccurate. To address this, we propose a spatial positioning method for magnetic mobile robots inside closed steel box girders, combining ultra-wideband ranging with structural boundary constraints. Distance measurements between anchors and a mobile tag are processed via a multispherical intersection algorithm, and the geometric boundaries of the girder are incorporated as hard constraints to optimize trajectory estimation using adaptive filtering. Experimental validation on an in-service bridge showed that static positioning errors can be controlled within 10 cm, while dynamic trajectory errors are significantly reduced after constraint optimization. The layout of anchors and robot speed were found to critically influence accuracy, with a recommended speed limit of 0.05 m s −1 for sub-decimeter precision. This method provides reliable, low-power positioning support for robotic inspection in complex steel structures.