Laser triangulation-based distance measurement using deep neural networks

This work aims to simplify and optimize the data processing of laser triangulation ranging method by utilizing deep learning for intelligent measurements, leading to enhanced measurement accuracy. Firstly, a nonlinear programming genetic algorithm with elitist strategy (E-NPGA) is employed to determine the optimal optical system parameters, resulting in the development of a laser triangulation ranging system. A deep neural network is then established for distance evaluation. Experimental results demonstrate that the deep learning based method proposed in this work enables a distance measurement with a root mean squared error within the range of 0.8 to 1 μm and a mean absolute error of below 0.7 μm, which offers a feasible intelligent measurement approach.