Performance Analysis of Neural Network-Based Uncalibrated Hand-Eye Coordination
Performance of a neural network-based control scheme is investigated for uncalibrated robotic hand-eye coordination system. Since the conditions for offline modelling with neural network are normally different from those for online control, unmodeled dynamics is inevitable and should be compensated by controller design. We analyze the system’s tracking error and stability with a discrete system model under a PI visual servoing controller, taking account of robot dynamics and image processing delays. The internal model principle is adopted to arrive at a feedforward compensator to enhance system performance.
KeywordsTracking Error Visual Servoing Root Locus Robot Hand Feedforward Controller
Unable to display preview. Download preview PDF.
- 1.Scheering, C., Kersting, B.: Uncalibrated hand-eye coordination with a Redundant Camera System. In: Proc. IEEE Inter. Conf. on Robot. Automa., pp. 2953–2958 (1998)Google Scholar
- 6.Su, J.: Dynamic Coordination of Uncalibrated Hand/eye Robotics System Based on Neural Network. Journal of System Engineering and Electronics 12, 45–50 (2001)Google Scholar