Gain Tuning Using Neural Networkfor Contact Force Control of Flexible Arm

Abstract

In this research, contact force control of an one link flexible arm is presented. A simple boundary feedback controller consisting of bending moment at the base of the flexible arm proposed by Endo et al. A gain adjustment control system using a neural network is designed and its control performance examined and compared by numerical simulation and experiment. In this study, we designed the feedback gain to correspond to the coupling coefficient of the neural network, and stabilized the learning by giving the initial value to the coupling coefficient of the neural network, thereby shortening the learning time. Also, in order to adjust the gain value in real time, a sequential correction type (online learning) that repeats learning at every sampling was adopted as the learning scheme of the neural network. As a result, it was confirmed that by using the using the neural network, the value of the feedback gain is adaptively changed and the target contact force converges around 0.35 seconds. Comparing with the fixed gain results, it takes shorter time for convergence to the target value by 0.8 seconds, the proposed controller is confirmed to be more effective for the contact force control of the flexible arm. Keywords—Flexible arm, contact force control, neural networks, gain tuning