Target tracking by visual servoing

Abstract

In the last few years, technological developments of vision sensors such as minaturization and strength have enabled their placement directly on the effectors or mobile robot manufacturers. The main advantage of such integration is to provide sensory information between the robot and its environment and hence open new prospects in robotics. In this paper, we propose to control a turret Pan-Tilt system using the visual features acquired by a webcam. These visual features are extracted by the correlation method and optimized by a multi scale pyramidal algorithm. The image based visual servoing scheme is used to compute the kinematics screw of the camera in the image frame. In order to use this kinematics screw in robot frame, we need to use the jacobian of the robot.