Robot stereo vision calibration based on hybrid swarm intelligent optimization

Accurate stereo vision model is the basis of robot high-precision visual positioning, however, it is difficult for the traditional or single non-linear optimization algorithm to achieve stable and high-precision calibration for robot stereo vision. Combining with strong global search ability of genetic algorithm (GA) and strong local search ability of particle swarm optimization (PSO), a three-step robot stereo vision calibration method based on hybrid swarm intelligent optimization is proposed. The calibration method is based on robot binary vision nonlinear model, linear initial values and first nonlinear optimized values of single camera models can be obtained in the first and the second steps individually, and the nonlinear optimization of stereo vision model are taken in the third step. Direct linear transformation, GA and PSO are individually used in three stages, and the result of every stage are used to initialize its next stage. Simulation analysis and actual experimental results indicate that this calibration method can work more accurately and robustly in noise environment, compared with other calibration methods using traditional optimization or single swarm intelligent optimization, and can better meet the requirements of robot sophisticated visual operation.