Research And Implementation Of Robot Visual Tracking Based On Particle Filters

As a challenging research topic in the field of robotics, visual tracking is widely applied to military, visual surveillance and digital medical treatment. Based on recursive Bayes theory, colored target tracking for mobile robot with monocular vision is discussed in this paper. A comprehensive research work is carried out in object detection and object tracking, which are central techniques in mobile robot's visual tracking in different environments. According to these tracking algorithms with different target models, mobile robot can accomplish the tracking tasks in both stationary and moving cases.A group of target detection methods are introduced and analyzed on the basis of color vision and colorimetry theory. Considering system's requirement in real-time processing and practical application, two detection algorithms based on weighted color histogram and run length encode are studied.The representation of Bayesian tracking is given based on Markov hypothesis and Bayesian formula, and two of its common implementation methods are introduced in succession. Aiming at the tracking task in non-Gaussian measurement environment, Monte Carlo Methods and particle filter are introduced in detail. In order to implement the real-time tracking on the mobile robot, a novel approach of hybrid particle filter algorithm is presented to process the target's position and shape respectively, whose states updating is on the basis of data fusion between Kalman filter and particle filter. The proposed method can not only pave the way for a low-complexity particle filter algorithm in dealing with higher dimensional tracking problem, but also cover the shortage of Gaussian restriction in Kalman filter.To ensure the effectiveness of object tracking when the robot is navigating, an object prediction method is proposed to reduce the impact caused by the movement of the camera, which is deduced from the camera's imaging process. Due to the variations of the targets' features, different target models, detection methods and tracking algorithms are employed to achieve successful tracking behaviors under various environments. A series of experiment results and data analysis show the method's validity and practicability.