Top-down image processing and supervisory control limitations in robotics: a simulation study

Abstract

A simulation study is used to evaluate the performance of an autonomous mobile robot system, with a visual feedback subsystem that employs a model-based top-down image processing algorithm. Visual redundancies are designed into the visual system for robust performance. Measures are designed to quantify the performance of the visual subsystem as well as the path tracking performance of the system under various degrees of simulated plant and image noise. Then the amount of visual redundancy in the visual system is varied so as to analyze the role of redundancies in overcoming the effects of the noise. Results showed that visual redundancy provided for more robust performance by the visual subsystem. However, the overall system performance can only take advantage of such improvements at high noise level. At low noise levels, the additional overhead actually has an adverse effect on system performance.