AquaNav - Unmanned Surface Vehicle Simulator Using Unreal Engine 5

Developing vision based algorithms for Unmanned Surface Vehicles (USVs) presents significant challenges due to the high cost and inherent risks of open ocean testing. As a result, simulation has become an essential tool in the early stages of development. In this context, we present AquaNav, a realistic simulator built on Unreal Engine 5. It features river, docks, and coastal environments with wave physics and real time lighting variations. AquaNav includes a surface vehicle which can be either teleoperated or made to follow certain desired velocity. The simulator supports the capture of co-registered RGB and depth images at configurable frame rates, and allows users to customize the simulated environment by selecting specific latitudes, longitudes, time zones, and time of day. This enables the generation of diverse and challenging visual datasets. The time of day can also be modified at runtime, providing a platform to evaluate the robustness of visual algorithms under varying environmental conditions. Additionally, we provide a preliminary labeled dataset generated using AquaNav. We also demonstrate sample vision algorithms operating while the simulation is running. The project is available at: https://github.com/AIRLabIISc/AquaNav.git