Vision-based cylinder posture recognition and on-line adjustment strategy

Vision-based micromanipulation robots are now widely used in the field of industry, military, biology, etc. for their minimized size, low energy consumption and reliable performance. This paper presents a cylinder´s axis extraction method for the foam column assembly system through image processing and feature recognition. A novel image Jacobian matrix estimation method based on Broyden update for uncalibrated visual servo system is proposed to adjust the posture of the foam column online, then its simulation model is established and tested by typical signals and the results demonstrate the effectiveness of the method.