BOND GRAPH MODEL-BASED INVERSION OF PLANAR PARALLEL MANIPULATOR SYSTEMS

摘要

Overwhelming controllers have been successfully used in the past for trajectory and impedance control. An overwhelming controller uses an image or copy of the system model in the modelbased controller. This model represents the kinematic constraints for implicit system inversion. The controller output is used to actuate the controlled system through a high feedback gain, which ensures that the dynamics of the controller overwhelms the plant dynamics. The overwhelming controller implementation for a serial robotic system is a straightforward task. However, implementation of a bond graph model-based overwhelming controller for a parallel manipulator shows up a few difficult causal loops. Removal of these causal loops and a few other subtle modifications yield a computationally efficient inverse model for real-time implementation. As validation of the developed approach, trajectory tracking in constrained planar welding operation is considered. The corresponding simulation results are presented.