Estimate the volume of residual frozen coal in railway carriage using a rotating LiDAR

To detect frozen coal residue in railcars during winter transportation in mid-to-high latitude regions, this paper proposes a method using a spin-type multi-line LiDAR system. The LiDAR is positioned above the railcar, with its rotation axis aligned to the train's direction, capturing point cloud data as the train moves. The data processing involves three steps: first, extracting point cloud data within the railcar's range based on the LiDAR-railcar positioning; then correcting for contour tilt and motion distortion and stitching the data using a motion displacement fusion algorithm. Statistical filtering and voxel grid methods are applied to filter, simplify, and smooth the stitched data. Finally, a 360-degree ray and alpha-blending algorithm extracts contour slices used to estimate frozen coal volume. Experiments were conducted to optimize voxel grid size and slice spacing parameters. With optimal configuration, the proposed technique achieves over 93.5% accuracy, addressing the inaccuracy of manual estimation and supporting frozen coal removal planning.