Some results on low resolution airborne ultrasonic image generation and processing

Abstract

The paper presents some preliminary results of airborne ultrasonic image generation and processing. The work relates to mobile robots which use artificial vision to navigate and to recognize objects from the environment. Ultrasonic images are generated by sonar heads. The images are obtained by a discrete approach, i.e. element by element. Two methods of building ultrasonic images are presented, both under biomimetic behavior of wild animals (bats): one emitter and two receivers. The first method uses a minimum number of processing functions in order to detect - as quickly as possible - the obstacle (size and shape) from the explored environment. The second method has - comparing with the first one - more processing functions and could be used to generate ultrasonic images for object recognition. Ultrasonic images have low resolution, i.e. 20 by 20 elements. The obtained results show that such low resolution ultrasonic images can be used for various navigation tasks, such as detection and recognition of objects.