Accurate light source position estimation for a laser triangulation measurement device using particle swarm optimization

The two-dimensional object's position or object's edge position can be well estimated with simple triangulation techniques utilizing the projected shadow of the object edge by exposing the edge in different angles. This can be done using two bare Laser diodes projecting the object's edge in a different angle onto a CCD or CMOS sensor without the need of any additional optical elements such as collimation lenses. For simple triangulation methods, sensor- and Laser diode position play a very important role. The overall accuracy is mostly determined by the uncertainty of these positions. This paper presents a possible approach to determine the light source positions highly accurate. Because deterministic methods suffer when badly shaped error functions occur, stochastic methods can deliver very good results, ending up very close to the global optimum. When using particle swarm optimization (PSO) for determining the light source positions, it could be shown that the overall 2D X/Y object's edge position estimation error, which strongly depends on those positions, can be significantly lowered. The 2D edge position estimation error can be reduced from 407 µm down to 3.4 µm in X direction and from 280 µm to 8.3 µm in Y direction, using a standard sensor and two bare Laser diodes. It was possible to show the good convergence of the proposed PSO method, firstly by setting up a simulation and testing with simulated edge data. Later, simulation results could be verified with real measurement data. When using PSO, it can be pointed out, that the light source position can be determined within micro-meter accuracy, which effectively reduces the overall measurement error of the system.