Overview
A simulation environment for testing autonomous underwater vehicle navigation and control before deploying to hardware, with an emphasis on realistic sensor and hydrodynamic modelling.
Architecture
Built on top of HoloOcean, the environment provides simulated DVL, IMU, depth, and imaging sonar sensors alongside a configurable hydrodynamic model of the vehicle hull.
Design
The simulation is treated as a first-class test environment rather than a demo: sensor noise, latency, and dropout are modelled explicitly so that navigation software is validated under realistic conditions.
Implementation
Vehicle dynamics and sensor models are configured through Python, allowing the same mission and navigation code used on hardware to run unmodified in simulation.
Experiments
Current experiments compare simulated DVL bottom-track behavior across varying altitudes and terrain types to validate sensor modelling fidelity.
Lessons Learned
Investing in simulation fidelity early paid off directly during the AUV project, where several navigation issues were caught before any hardware testing occurred.