摘要：For the eye-to-hand monocular vision system, based on the idea of "black box", which maps the image coordinates directly to the robot reference coordinates, the transform matrix about image coordinate and robot reference coordinate is directly obtained in the condition of fixed distance between the positioning plane and camera. The optimized transform matrix can be obtained as global calibration parameters for the eye-to-hand system, without the need for calculation of camera’s internal and external parameters. As the transformation matrix does not consider the non-linear imaging of the lens, the direct use of this transform matrix of the two coordinates system will have a greater accuracy error. In order to reduce the positioning error of hand-eye vision system, this paper considers the obtainment of transform matrix as multiobjective optimization problem and applies the hybrid particle swarm algorithm to optimize the matrix parameters. The experimental results show that this method can effectively improve the hand-eye positioning accuracy and can be successfully applied to industrial production.
