Robust model-based tracking for robot vision

摘要: This paper proposes a real-time, robust and efficient 3D model-based tracking algorithm for visual servoing. A virtual servoing approach is used monocular tracking. method similar to more classical nonlinear pose computation techniques. concise derivation of distance-to-contour interaction matrices described. An oriented edge detector in order provide real-time points normal the object contours. Robustness obtained by integrating M-estimator into control law via an iteratively reweighted least squares implementation. The presented this has been validated on several 2D 1/2 experiments considering various objects. Results show be occlusion, changes illumination miss-tracking.