Study Of Target Tracking And Obstacle Avoidance For Mobile Robot Using Kinect

Robotic technology is a subject which includes multi-disciplinary field. It integrated computer vision, kinematics control, environmental information perception and other subjects.As robotics hot topic,target tracking and obstacle avoidance technology of mobile robot have widely been paid attention. It is very necessary for robot to take advantage of mobile robot sensors environmental information of moving target tracking. In contains obstacles environment planning out a the best route to the target point. There for target tracking and obstacle avoidance method has theoretical and realistic significance. In this paper, the main research contents and results are presented as follows:(1) In the indoor environment, Kinect were used as a mobile robot vision sensor, color image and depth image were obtained through Kinect. Cover the shortage in the process of the traditional image processing by using single color information is insufficient, Laid a good foundationfor the mobile robot moving target tracking and avoidance.(2) In this paper, A tracking algorithm based on the detection of the tracking algorithm--TLD(Tracking-Learning-Detection) were used. TLD algorithm is a algorithm of single target and long time tracking Detailed analysis of the TLD tracking algorithm to track, study, testing in three parts. TLD part tracker algorithm is based on a forward-backward LK optical flow. Optical flow method in practical applications. Since the influence of multi-source, blocking resistance,transparency and noise. The basic equations of the flow field of the light gray Conservation assumptions are not met, can not solve the correct optical flow field, while the optical flow calculation method is very complex and can not meet the real-time requirements. So using a smaller amount of calculation algorithm replaces the original multi-instance TLD tracking optical flow section, a multi exemplary TLD tracking algorithm based on the integration of depth information. Experimental results show that the algorithm can effectively track changes in the target posture, occlusion, etc., and in the case of similar target color by interference, but also to get a better tracking.(3) In this paper, aimed at the shortage of a narrow channel capacity is deteriorated,artificial potential field method to increase security in the region path planning and obstacle avoidance. Kinect were used to access environmental information and stored as a map to facilitate the use of mobile robots. The local map update method were come up due to the Kinect blind spot problem Experimental results show that the proposed algorithm can successfully bypass the mobile robot stationary obstacles in unknown environment and tracking moving targets, also improve the ability that the mobile robot along the wall and go through the narrow channel.