Robot Vision Using Cellular Neural Networks

摘要

We show how Cellular Neural Networks (CNNs) can provide the necessary image processing to guide an autonomous mobile robot in a maze made of black lines on a light surface. The system consists of a fuzzy controller performing the elementary navigation tasks fed by the result of processing the image only by CNN techniques. We use this solution to make some considerations on more difficult problems such as curved or dashed line following and obstacle avoidance.