An Economic Simultaneous Localization and Mapping System for Remote Mobile Robot Using SONAR and an Innovative AI Algorithm

Abstract

Abstract-This paper proposes an economic and effective approach towards the simultaneous localization and mapping of a mobile rescue robot using a single ultrasonic range finder. The procedure eliminates the complication involved with localizing the robot in a map while creating the map simultaneously, by employing a novel control mechanism. The problem is solved by separating the mapping and localization processes and merging the outputs after specific intervals. The data from sensory devices is processed wirelessly to map the surroundings of the robot on a computer. The innovative AI algorithm extends the map cognitively as subsequent data is acquired. Moreover, a comparison is made between conventional methodologies and the one presented in this paper along with possible enhancements applicable. Using this architecture, the economic cost, computational power and time required for SLAM are significantly reduced. The design was utilized on a mobile rescue robot for the RoboCup competitions.

Research on sonar pulse recognition based on convolutional neural network

“…In the field of underwater acoustic target recognition, sonar pulse signal recognition is a key problem to be solved by the underwater battlefield environment monitoring system, which is also an important criterion for evaluating combat decisions on the underwater battlefield. [1][2][3][4][5] In recent years, the application field of artificial intelligence technology has been expanding, and in terms of related radar signal processing, domestic and foreign scholars mainly use supervised learning methods [6] to extract inherent feature parameters, train classification models, and predict image types with the help of unlabeled signals. Jing Bojun et al recognized radar signals and the recognition rate of multi-type signals reached 92% [7] .…”