Target tracking for mobile robot based on Spatio-Temporal Context model

Object tracking is essential as a basis in robot field. In this paper, a tracking system based on the Spatio-Temporal Context (STC) model is proposed to make most use of effective information between the appearances of the target and the changes of its surroundings over space and time. The platform is by necessity equipped with a Microsoft Kinect which is used to capture the images of the scene and obtain the position of target. Then, the depth information is used for the robot localization and target tracking. Based on the information of the target, fuzzy-based intelligent control strategy is applied to adjust the movement of the mobile robot according to the target´s speed and the distance from the robot. It contains two fuzzy controllers for adjusting the linear velocity and the turning-gain. Finally, the experimental results conducted on the Pioeer3-DX show that it can follow the target robustly and have a good real-time performance.