Single-beam Scanning Imaging Sonar Data Processing And Implementation By DSP

The research of AUV has brought many new opportunities for human to explore and develop the oceans. At the same time it has also brought big challenges to the worldwide robot research, the greatest of which is how to achieve fully autonomous navigation of underwater robot. The possibility of robot's autonomous navigation depends on autonomous navigation methods, such as SLAM (Simultaneous Localization and Mapping). Whether by using any ways of navigation, it is an indispensable step to obtain the background information from the detecting environment where the AUV locates, which is the premise to achieve precise localization and motion control of underwater robot. During underwater environment, sonar has incomparable superiority than other equipments.In this paper, the single-beam scanning sonar is used as main sensor to detect the surrounding environment of AUV to collect environmental information and create the acoustic images. In combination with sensors including DVL, Digital Compass and Gyroscope which measure and collect the movement data of AUV, using SLAM (Simultaneous Localization and Mapping) as the navigation method, the autonomous localization and navigation of underwater robot can be achieved.In this paper, after the full study about the advantages and disadvantages of kinds of sonar data processing methods, considering the technical features and working principle of the single-beam scanning imaging sonar, and in accordance with the necessities of robot autonomous navigation method, a sonar data processing method is presented, which is real-time and has the ability to extract accurate point features, then the concrete realization is achieved through a sonar data processing system with C6000 DSP as the CPU core. During the real-time sonar data transmission process, a data preprocessing is carried out through data separation and threshold segmentation; then through single-ping echo data sparse processing and the entire round echo data sparse processing, the sparse point features are extracted. Then two kinds of experimental sonar data are used to compare the results of point features extraction. The extracted point features will be available for navigation system to do estimation of robot's pose and creation of environmental features map. This method is real-time and has the ability to extract accurate point features, so it is particularly suitable for the autonomous navigation of underwater autonomous robot.