| The sensing capability of robots must be improved when they do some dangerous works in dynamic changing environment. One way to improve the sensing capability of robots is to mount all kinds of sensors on the robots, for example, force sensor, haptic sensor, proximity sensor, range finder and vision sensor.As far as the autonomous robot is concerned, the vision sensor plays animportant role because it can provide the noncontact measurement of thechanging environment and get rich information about the changingenvironment. With the help of data from other sensors, the robot can make adecision, run corresponding operation, and finish the task. The vision sensor iscombined in open-loop fashion traditionally and used to measure the positionand orientation (i.e. pose). The robot implements corresponding operation interm of the measured pose. The operation accuracy depends completely on theaccuracy of vision sensor and the manipulator and its controller. When theobject changes it position during the robot operation, the task will be fail.Placing the vision sensor in the closed-loop of robot controller and using thevisual information to control the pose of the robot (for example, manipulator,vehicle, flying robot and underwater vehicle) with respect to the target objectis so called visual servoing. Visual servoing uses the visual information toguide the robot to implement the object tracking, grasping and assembling, andcan improve the accuracy of operation and the autonomous capability of robot.According to how the visual information are used in the feedback loop,Visual servoing schemes are classified into three groups, namely,position-based, image-based and 2.5D visual servoing. According to theposition that the camera mounted, visual servoing schemes are classed intoEye-in-Hand and Eye-to-Hand configuration. According to the number of thecamera mounted in the system, visual servoing schemes are grouped intomonocular, binocular and multi-camera configuration visual servoing system.
|
PDF Full Text Request