Research on vision-based robot planar motion measurement method

Abstract

The dynamic and static characteristic detection of robot is of great significance to estimate the performance indices of robots and evaluate the performance of robots in the processes of robot design manufacturing, integrated application and maintenance. The dynamic and static characteristics, such as control accuracy, bearing capacity, deformation, vibration and structural mode, can be calculated by measuring the motion of the robot, which can be converted into corresponding index parameters to evaluate the behaviors of the robot. Hence, it has always been a central research topic in the field of robots to realize robot motion measurement rapidly and accurately. In this paper, a vision-based robot planar motion measurement method based on the homography characteristic matrix was proposed. Using this method, the problems, for instance, of poor precision, repeatability and flexibility existing in the scale factor approach, which is widely used in similar studies due to its simple mathematical model, are solved. To further improve the measurement accuracy and stability, the singular value decomposition method of the characteristic matrix was drawn into solving the model parameters on the basis of the traditional direct linear transformation parameter calibration method, and a novel feature point extraction method was proposed to acquire sub-pixel precision features. Several experiments were carried out on a self-built three-degree-of-freedom rectangular coordinate robot platform on which the marker is glued. The experiment results show that the aforementioned method can successfully reconstruct the planar motion of a robot accurately, which can satisfy the requirements of high-precision motion control, motion performance evaluation and operation state evaluation.