Abstract
Underwater remotely operated vehicles are widely used in various marine-related tasks. This ranges from environmental studies to underwater structure assessment. With the increasing use of AI-based solutions, there is a need for improving the existing compute of underwater vehicles. In this paper, we propose a Modular, Adaptable Research Vehicle Loadout for the BlueROV2 (MARVL-ROV). The MARVL-ROV includes the upgrade schematics for attaching an onboard edge computing alongside various sensors such as the imaging sonar and the doppler velocity logger. MARVL-ROV also has a removable tether for flexibility in performing both remote-operated tasks and autonomous tasks. We describe the additional designs made to account for changes in buoyancy and communication requirements. Finally, we implemented autonomous controls to test the capabilities of MARVL-ROV without tether and present our findings. A living documentation of MARVL-ROV is made available at GitHub:MARVL-Lab/MARVL-ROV.