Research And Design On The Underwater SLAM In Man-Made Structured Environments Based-on Multi Single-beam Sonars

Simultaneous localization and mapping (SLAM) is one of the fundamental problems that need to be solved before achieving truly autonomous vehicles. For this reason, in recent years it has been the focus of a great deal of attention. Multiple techniques have shown promising results in a variety of different applications and scenarios. Some of them perform SLAM indoors, outdoors, and even on air. However, the underwater environment is still one of the most challenging scenarios for SLAM because of the reduced sensorial possibilities and the difficulty in finding reliable features.This paper describes a navigation system for unmanned underwater vehicles (UUVs) in partially structured environments, such as dams, harbors, marinas, and marine platforms. The position estimate was achieved through a Suboptimal fading extended Kalman filter (SFEKF)-based dead reckoning. The dead reckoning used the data from all sensors (including the multi-single-beam sonars) and the nonlinear dynamic model of the UUVs.Hough Transform has been developed to extract line features, together with their uncertainty, from the continuous multi-single-beam sonars data flow. The obtained information is incorporated into a feature-based simultaneous localization and mapping (SLAM) algorithm running a Suboptimal fading extended Kalman filter. Simultaneously, the UUV's position estimate is provided to the feature extraction algorithm to correct the distortions that the vehicle motion produces in the multi-single-beam sonars data.A data set obtained during an experiment performed with the Bai Tun UUV serves as a test for the proposed SLAM algorithm. The vehicle performed a 1400-m trajectory in an abandoned marina situated in the Xiao Ping Island in Da Lian. The resulting data set includes measurements from many different sensors. Multi-single-beam sonars provide range data that feeds the feature extraction algorithm. The velocity measurements from a Doppler velocity log (DVL) are introduced in the filter to perform dead reckoning, and OCTANS and a pressure sensor provide absolute measurements for the vehicle heading and depth.We demonstrate the practicality of this approach by building a geometric map of a large size, real marina environment. The marina-experiment results verified the validity of this approach. Maps built from GPS data for the same experiment are provided for comparison.