Improving Environmental Perception During Underwater Manipulation Through Stereo Vision and Acoustic Imaging: A Simulation-Based Approach

Performing precise underwater manipulation is a challenging task, especially when applied to the field of underwater cultural heritage (UCH). In fact, errors in measuring the distance between a predefined target and the manipulator's end-effector can lead to significant issues, such as collisions or challenges in grasping objects. Therefore, there is a need to equip operators with tools that enhance their visualization and interaction capabilities, enabling more precise underwater manipulation. To this end, the paper introduces a novel simulation-based approach to increase environmental perception during underwater manipulation operations based on the adoption of stereo vision and acoustic imaging sensors. Specifically, the paper presents the software architecture of an underwater simulator that leverages a graphics engine to replicate a system comprising a Remotely Operated Vehicle (ROV) equipped with a robotic manipulator, stereo camera, and forward-looking sonar. The proposed simulation framework leverages both the Unity engine and ROS2 to enhance underwater robotic simulations. This paper presents preliminary results on stereovision and acoustic simulation, demonstrating how the integration of these sensing modalities improves environmental perception during underwater manipulation tasks.