Testing & Validation Pipeline
To ensure the reliability of our AUV, MM Nautronics implements a rigorous, multi-phase testing strategy. By decoupling software validation from hardware readiness, we maximized our development speed and secured over 8 months of active in-water testing. Below is the summary of our testing events, objectives, and critical engineering results.
Phase 1: Virtual Simulation & FEA/CFD
Duration:
3 Months (November 2025 – January 2026)
In this phase, the vehicle was tested in a digital environment using Gazebo, SimScale, and Autodesk CFD. Before manufacturing, hydrodynamic efficiency, structural strength limits, and autonomous decision-making processes were evaluated virtually. This allowed design and software decisions to be validated in a risk-free environment before moving on to physical testing.
Pressure-induced displacement (deformation) analysis of the main vehicle
Pressure-induced displacement (deformation) analysis of the main vehicle
Steady-state CFD analysis on the thruster propeller
CFD analysis illustrating flow streamlines and velocity magnitude around the thruster
Thermal analysis results of the electronics organizer
Virtual simulation test environment
Key Results
Hydrodynamic Optimization
Steady-state CFD analysis showed a massive cavitation risk (-55.5 MPa local vacuum) if we pushed the thrusters to 2000 RPM. Result: We capped the operational speed to the 1000-1500 RPM sweet spot. This simple tweak saved 60-70% in battery consumption, allowing us to drop a bulky 6S LiPo in favor of a compact 4S LiPo.
Structural Integrity
We simulated extreme pressure at a 25-meter depth to see if our sealing strategy would hold. Result: The main PMMA tube deflected by only 0.163 mm. This successfully validated our O-ring compression tolerances without risking a physical implosion.
Software & Logic
Hardware wasn't ready, but our software team couldn't wait. Result: We regression-tested our entire 'py_trees' Behavior Tree logic in Gazebo, ensuring the vehicle's sequential state memory worked flawlessly before it ever touched the water.
Phase 2: Laboratory & Bench Testing
Duration:
3 Months (January 2026 – March 2026)
In this phase, the vehicle’s mechanical, sealing, and electronic subsystems were tested in laboratory and workshop conditions before being placed in water. The dry testing process allowed potential mechanical mismatches, connection issues, sealing risks, and electronic failure conditions to be identified at an early stage. This enabled the reliability of the subsystems to be evaluated before moving on to wet integration.
Laboratory (job-shop) Test Environment
Laboratory and workshop test environment where the mechanical, electronic, and sealing subsystems were validated before being placed in water.
Key Results
Mechanical & Actuator Systems
We ran repeated dry-cycling tests on our servo-driven subsystems—specifically the grabber, marker dropper, and torpedo launcher—to confirm correct actuation timing and mechanical consistency before any water exposure. Also, during bench stress tests, the original HDPE thruster mounts fractured under load, prompting us to redesign and manufacture them using PETG for superior tensile strength.
Sealing Integrity
Initial vacuum tests were insufficient for detecting capillary micro-leaks around the penetrators. Result: We overhauled our protocol by introducing a 150-psi "Positive Pressure Bubble Leak Test," allowing us to visually pinpoint and patch defects before entering the water.
Electronic & Thermal Validation
The fully integrated electronics organizer was bench-tested under continuous full load. Result: Transient thermal analysis confirmed that ESC temperatures safely plateaued at 53°C, successfully validating our Aluminum 6061 end-caps as effective passive heat sinks.
Phase 3: In-Water Field Testing
Duration:
8+ Months Total (Parallel legacy AUV testing + New AUV integration)
In this phase, the AUV’s performance was evaluated under real in-water operating conditions. A pseudo-competition test setup was created at the METU pool facility to test full system integration, buoyancy balance, and the vision pipeline. Through parallel testing on the legacy AUV and the integration process of the new AUV, the team gained more than 8 months of in-water testing experience.
In-water pool test environment
Key Results
Vision Pipeline Tuning
Early pool sessions proved that underwater light refraction and turbidity easily confused our YOLOv8 model. Result: We countered this by integrating hardware color filters and OpenCV-based pixelation, which dramatically stabilized our object detection.
Sub-system & Actuator Integration
We moved our isolated actuator tests into the pool to validate underwater deployment. Result: We successfully calibrated the marker dropper under real hydrodynamic drag, ensuring that the vehicle’s attitude control loop instantly counteracts the sudden torque and buoyancy shifts during actuation.
Software Architecture
High-frequency ROS2 communication started bottlenecking the Jetson CPU during complex maneuvers. Result: We moved the heavy processing to C++ and linked it with Python via pybind11. This hybrid setup gave us the raw speed of C++ with the flexibility of Python, completely eliminating the communication lag.
Buoyancy & Dynamics
We needed to physically verify our Archimedes calculations in the pool. Result: The vehicle settled at a perfect 3.03% positive buoyancy. By deliberately lowering the center of gravity, we achieved a self-righting pendular effect that naturally fights off wave disturbances.