Pipeline inner surface 3D vision measuring system based on robot equipped with multi-directional structured-light sensor

Abstract Vision perception and precise measurement are crucial for enhancing the detection capabilities of pipeline robots. The structured-light (SL) sensor is widely used in pipeline inspection due to its strong resistance to interference and high stability. In order to implement automatic inspection of a pipeline’s inner surface, this paper develops a three-dimensional (3D) vision measuring system based on a robot equipped with multi-directional SL sensor. The geometric imaging model for the multi-directional SL sensor is firstly established, allowing quantitative analysis and simulation of the sensor’s optical imaging characteristics, which implement the optimization of structure parameters. The proposed sensor employs catadioptric components and line-SL measurement principles to achieve multi-directional vision measurement, offering greater flexibility compared to ring-SL method. Calibration method for the SL sensor parameters is proposed based on the SL local measurement model and multi-virtual viewpoint global measurement model. To improve the accuracy of 3D reconstruction of the pipeline, a point cloud reconstruction method is further developed based on coordinate reference transfer and optimization. Lastly, experiments are conducted to validate the accuracy of sensor calibration and pipeline diameter measurement, with an evaluation of the measurement uncertainty.