Pose detection of parallel manipulator based on visual tracking

This study addresses the pose detection of parallel manipulator based on visual tracking. Firstly, an improved mean-shift is used to track the moving platform, and then four points with significant features are chosen. At last, two affine invariants are exploited to detect the pose of the motion platform. The experimental results show that the parallel manipulator is tracked better by the proposed method under the condition of illumination changing obviously, and the pose parameters are obtained simply by the two affine invariants.