End-effector tracking in the micro-manipulation system with trajectory-aided particle swarm optimization

This paper presents a trajectory-aided particle swarm optimization (PSO) for tracking the end-effector in the micro-manipulation system. When the micro-environment suffers from significant interference with the result that the end-effector cannot be detected by micro-vision, failure of target tracking is often caused. Hence, the proposed method is developed to solve this problem by predicting the trajectory of the end-effector. A series of end-effector paths is firstly summarized. According to the Bayesian theory, the tracking problem is then converted to the problem of calculating the probability of motion paths when the loss of end-effector occurred. A cost function is established to determine the probability by optimization, and a decision criterion is established to judge whether the loss of target occurred. After that, the predicted trajectory is used to distribute particles. The proposed method preserves the advantages of PSO, such as less sensitivity to the initial conditions and high tolerance of the target model. Experiments with real micro-operation video for tracking the needle in different scenarios show that the proposed method greatly improves the tracking performance and time efficiency compared with the traditional PSO.