Conventional industrial robot systems can only repeat pre-programmed tasks. The systems cannot complete the tasks when any changes in their working environments occur. To cope with the unexpected changes, a robot system needs to sense the environment by visual sensors.
Visual sensory feedback can enhance the performance of a robot system in many applications such as automatic assembly,product processing and material handling. Visual servoing is especially important to deal with unknown environments and unexpected events for robot systems.
In this paper, we develop visual servoing techniques for a X-Y robot to track an arbitrary path on a plane.
First, we develop a conventional method based on camera calibration to track an arbitrary path. But this method shows time delay effects and poor accuracy because of image processing time delay and inaccurate robotic calibration.
Second, we present an adaptive visual servoing algorithm based on parameter estimation method. Since the adaptive algorithm predicts a desired path for the robot to track, the time delay and inaccurate calibration problems are solved.
Third, we propose another visual servoing algorithm based on on-line calibration. This algorithm performs a calibration during each tracking step without pre-calibration.
We evaluate and compare three approaches through a series of simulation and real experiments. The experimental results demonstrate that the proposed algorithms are less sensitive to the inaccurate robotic calibration and image processing time delay than the conventional method.
