Support vector regression learning based uncalibrated visual servoing control for 3D motion tracking

This paper proposed a new method to control the uncalibrated visual servoing for 3D motion tracking. Firstly, PI control based movement planning is employed in image plane. Then, support vector regression (SVR) is used to construct the visual mapping model. Finally, a flat and three-dimensional space motion tracking is achieved via using real-time motion planning. Compared with the 3D motion visual tracking with the traditional BP neural network method, the experimental results demonstrated that the SVR had an excellent approximating capability under the condition of small sample learning.