Visual Tracking of Robotic Manipulator Motion Trajectory Using Dynamic Active Contour Model

Abstract

A dynamic active contour model for spatio-temporal trajectory is proposed to track robotic manipulator motion in image sequence space.The model employs a snake function with minimum energy at the trajactory distribution.The manipulator is localized through searching the converged energy point set.Energy weights in the function are real-time adjusted to avoid local minima during convergence.To improve the snake searching efficiency,a quadratic-trajectory least square estimator is developed to predict motion position before tracking.Experimental results of tracking micromanipulator motion demonstrate the effectiveness of the proposed model.