DSP based uncalibrated visual servoing for a 3-DOF robot manipulator

This paper presents a DSP based online camera calibration procedure when camera position is changed. It is known that classical image based visual servo control scheme has some problems, e.g., image space singularities and local minima. To tackle this issue this study applied a virtual composite camera model and a 3D visual Cartesian space to calculate the full-rank image Jacobian for visual tracking control. The advantage of this approach is that the image Jacobian obtained will be always non-singular for robustness. This research employed a distributed embedded system including two DSP boards to implement the above visual servo algorithm. The experimental results demonstrate that the proposed visual servo system is feasible in an embedded system.