Intelligent object feature identification using a mobile sonar rangefinder

We are immersed in a high-tech computing environment, depending on its support to carry out our everyday tasks. Unfortunately, our rapidly increasing expectations of the intelligence of modem devices exceeds their abilities. We often expect devices to perceive their environment, understand our intent, and autonomously carry out appropriate tasks. This paper presents a hybrid genetic-fuzzy approach to bring device perception closer to our expectations. Mobile sonar rangefinders are selected for this approach to provide direct, timely measurements of the environment without the computational overhead required to extract depth information from vision based sources. Experimental results confirm the feasibility of this approach to intelligently identify planar object features. Benefits over traditional probabilistic approaches include the simplicity and flexibility of integrating additional sources of information, and the ability to identify planar object features without predetermined probability distributions.