Target tracking control of mobile robot in diversified manoeuvre modes with a low cost embedded vision system

Abstract

PurposeTarget tracking systems are generally computationally intensive and require expensive and power‐hungry visual sensors. On the other hand, the existing target tracking control approaches fail to track the target swiftly and accurately when the mobile robot moves in the diversified manoeuvre modes. The purpose of this paper is to propose a novel target tracking control method with a low cost embedded vision system to achieve high accuracy and speediness of target tracking control, regardless of the type of manoeuvre modes.Design/methodology/approachThe pan/tilt angle differences are transformed from the tracking error between the image centre and the coordinates of the target centroid returned by the CMUcam3; the corresponding pan/tilt angle variation rates are calculated based on the manoeuvre control. All of them are fed to the controller. Then the controller generates appropriate control signals to fit the changing speed of target centroid and compensate for the tracking error. The experiments are designed in a way that the CMUcam3 keeps the target centre coincident with the image centre when the mobile robot moves in the diversified manoeuvre modes.FindingsIn spite of the type of manoeuvre modes, the controller responds to the tracking error instantly and actuates the pan/tilt with suitable position and speed commands, and the target centroid remains in the bounding box during the entire movement.Originality/valueThe proposed target tracking control takes the correlation between the robot manoeuvre modes and the target tracking control into account, and particularly suits for the target tracking tasks in planetary exploration, surveillance and military applications.