SIO-UV: Rapid and Robust Sonar Intertial Odometry for Underwater Vehicles

We propose a sonar-inertial odometry framework named sonar-inertial odometry for underwater vehicles (SIO-UV) for underwater vehicles, designed to address the challenges posed by complex underwater environments and sensor limitations. The aim is to achieve efficient and robust simultaneous localization and mapping (SLAM) in underwater conditions. SIO-UV integrates observation data from an inertial measurement unit (IMU) and a forward-looking multibeam sonar (FLS) to construct factor graph nodes, using IMU preintegration to estimate the trajectory of the underwater vehicle and to provide an initial pose optimization for sonar observations. To enhance the robustness of sonar data, this method is based on the FLS model and incorporates a multiscale constant false alarm rate (MCFAR) denoising algorithm. It can convert 2-D sonar images into 3-D point cloud data. We introduce a reverse regression mapping method for FLS. Ultimately, the generated sonar point cloud is used to estimate sonar odometry and provide bias correction for IMU preintegration. Extensive validation of underwater mapping was conducted in advanced simulation environments, with quantitative experimental results demonstrating the high accuracy of SIO-UV in underwater vehicle odometry estimation. Furthermore, SIO-UV was successfully deployed in real-world experiments with a remotely operated vehicle (ROV), demonstrating its outstanding performance in real-time processing, robustness, and accuracy.