Hand-eye calibration of measurement robot based on multi-population particle swarm optimization

A robot can be used to non-contact measurement with a structured-light scanner fixed to its end effector. In order to transform the 2-dimension values measured by the sensor into the 3-dimension values in the world frame, the geometric model of the system is established. A method to estimate hand-eye matrix is presented, which uses a radius-known ball, whose center is derived from the least square fitting method to enable the sensor to measure a fixed point (ball center) from different views. Thus hand-eye calibration is converted to unconstraint optimization problem. A multi-population particle swarm optimization (MPSO) is used to search hand-eye matrix, which merges the advantages of the global PSO and local PSO. Simulation and experiment show the ability of the algorithm.