Performance of a memory-based approach to the control of robotic manipulators

The primary objective of this study is to examine the effectiveness of a memory-based control strategy for robotic manipulators. The main idea behind the memory-based control method is to use certain gathered information, such as most recent tracking error, current tracking error and past control experience, to directly modify the current control command. Simulations on a 3 DOF robotic arm show that this method is very effective in dealing with system nonlinearities and parameter uncertainties. Comparisons between the memory-based control and the existing model-based control are also provided.