Ray-Based Physical Modeling and Simulation of Multibeam Sonar for Underwater Robotics in ROS-Gazebo Framework

While sonar sensors are crucial for underwater robotics perception, the key challenge lies in traditional multibeam sonar simulation's lack of comprehensive physics-based interaction models. Such missing physical aspects lead to sonar imagery discrepancies, such as the absence of coherent imaging systems and speckle noise effects exposing risks of over-fitted control designs of the systems using the sonar perceptions. Previous research addressed this gap by introducing a physics-based simulation approach by direct calculation of the point-scattering model equations from perception data obtained from rasterization. However, the raster-based method could not control the resolution of data to pipeline into image generation, and its limitation was explicitly presented in local search scenarios where the distance between data is large. To eliminate those limitations and extend capabilities without losing the quality of the image, this paper introduces a ray-based approach to replace the raster-based method when obtaining the perception data from the simulated world to pipeline into physical equation calculations. The results of the ray-based and raster-based models are compared for the front floating object and the ground grazing local search scenario to confirm that the ray-based method maintains equal quality of sonar image generation, including physical characteristics, but it has more flexibility and capability in control of data resolution for correct sonar image generation.