On Adaptive Control for AGV with Vision Sensor as an Unknown Nonlinear System

Abstract

Adaptive control using neural network provides a real-time systematic approach to achieve or maintain a desired level of control system performance when dynamics are unknown. The paper proposes a novel approach for designing of an adaptive controller with input as relative pixel density from a fuzzy system for automated guided vehicle with vision sensor. The fuzzy system computes relative pixel density from vision sensor data while minimizing uncertainties due to illumination, occlusion, and obscure images. It provides a methodology to apply an advanced nonlinear intelligent control technique for vision-based path tracking problem. The proposed strategy has been applied for path tracking problem on an indigenously developed vehicle. The results obtained show the efficiency of proposed approach and ease of applying different control techniques for vision sensor-based plant.