Synergized Navigation Autonomy
A high-fidelity Unity-based marine vessel simulation platform for autonomy research, reinforcement learning, and COLREGS-compliant navigation development.
NaviSense Simulator is a research-grade digital twin for marine vessels — built for the engineers, scientists, and institutions developing the next generation of autonomous maritime systems.
It bridges Unity's real-time physics engine with Python's AI/ML ecosystem over a live TCP connection, enabling bidirectional control, sensor streaming, and reinforcement learning training loops — all in a physically accurate ocean environment.
Built for:
- 🔬 Autonomy researchers needing a realistic sim-to-real testbed
- 🤖 RL engineers building COLREGS-compliant navigation policies
- 🏢 Defense & maritime organizations evaluating autonomous vessel behavior
- 📡 Sensor fusion teams requiring noisy, physics-based GPS, IMU, and camera data
┌─────────────────────────────────────────────┐
│ Unity Simulation │
│ │
│ ┌──────────────┐ ┌────────────────────┐ │
│ │ Hydrostatics │ │ Wave Physics │ │
│ │ Controller │ │ (Crest Ocean) │ │
│ └──────┬───────┘ └────────┬───────────┘ │
│ │ │ │
│ ┌──────▼────────────────────▼───────────┐ │
│ │ ActuatorController │ │
│ │ Propulsion · Rudder · Bow Thruster │ │
│ └──────────────────┬────────────────────┘ │
│ │ │
│ ┌──────────────────▼────────────────────┐ │
│ │ PythonBridgeManager │ │
│ │ TCP :5005 · JSON · 5Hz streaming │ │
│ └──────────────────┬────────────────────┘ │
└─────────────────────┼───────────────────────┘
│ TCP Bidirectional
┌─────────────────────▼───────────────────────┐
│ Python AI Stack │
│ │
│ ┌────────────┐ ┌───────────┐ ┌────────┐ │
│ │ Gymnasium │ │ Stable │ │PyTorch │ │
│ │ Custom Env │ │ Baselines3│ │ Models │ │
│ └────────────┘ └───────────┘ └────────┘ │
└─────────────────────────────────────────────┘
Ownership Model:
- Python owns: X/Z translation, yaw heading
- Unity owns: heave (Y), roll, pitch — governed by live hydrostatics physics
| System | Description |
|---|---|
| TCP Bridge | Bidirectional JSON protocol over TCP port 5005; 5Hz sensor stream, camera JPEG frames |
| Hydrostatics Controller | Buoyancy, heave spring/damping, pitch/roll restoring moments with multi-strip wave sampling |
| GPS Sensor Model | White noise + first-order Gauss-Markov bias (60s τ, 1.5m σ) + Bernoulli dropout |
| Actuator System | PropulsionActuator, RudderActuator, BowThrusterActuator with LocalDynamics/ExternalState modes |
| Wave Physics | Crest Ocean integration — least-squares slope estimation for pitch/roll excitation |
| Vessel Configuration | ScriptableObject-based vessel asset system (366t DolphinExplorer reference vessel) |
| System | Status |
|---|---|
| SimulationManager | Stub — session lifecycle, pause/resume, state machine (MVP priority) |
| ScenarioManager | Stub — scenario loading, multi-vessel spawning, COLREGS test cases |
| BridgeManager | Stub — unified bridge abstraction over PythonBridgeManager |
| System | Description |
|---|---|
| RL Training Loop | Gymnasium environment wrapper, PPO/SAC training pipeline |
| COLREGS Engine | Rule-based encounter detection (head-on, crossing, overtaking) |
| Replay System | Record/playback scenario runs for evaluation and demonstration |
| REST API / SDK | External integration layer for third-party research tools |
◉ M1 — Core Engine Stabilization [Active] Weeks 1-3
├─ SimulationManager implementation
├─ ScenarioManager implementation
└─ Integration test suite
○ M2 — Python RL Integration [Planned] Weeks 4-6
├─ Gymnasium custom environment
├─ PPO baseline training run
└─ COLREGS encounter detection
○ M3 — Data & Evaluation Pipeline [Planned] Weeks 7-8
├─ Synthetic dataset generation
├─ Scenario replay system
└─ Performance benchmarks
○ M4 — MVP Release & Documentation [Planned] Weeks 9-10
├─ SDK / API documentation
├─ Demo scenario package
└─ Research paper draft
Target MVP: Q3 2026
Prerequisites: Unity 2022.3 LTS · Python 3.10+ · Crest Ocean System (Asset Store)
# Clone the repository
git clone https://github.com/YOUR_USERNAME/navisense-simulator.git
cd navisense-simulator
# Install Python dependencies
pip install -r requirements.txt
# Open the Unity project
# File → Open Project → select the project folder
# Load scene: Assets/Scenes/SimulatorBase.unity
# Start the Python bridge
python bridge/main.pyNote: Full setup documentation coming in M4. For early access or research collaboration, see contact below.
Research & Development
- Train RL agents on physically accurate vessel dynamics without hardware risk
- Generate large-scale synthetic sensor datasets for perception model training
- Benchmark autonomy algorithms against standardized COLREGS scenarios
Defense & Maritime Industry
- Evaluate autonomous vessel behavior in controlled simulation before sea trials
- Test sensor fusion pipelines under realistic noise and degradation conditions
- Prototype guidance and control systems prior to hardware integration
NaviSyn Marine Solutions is actively pursuing SBIR/STTR funding through Navy and DoD programs focused on maritime autonomy. We welcome collaboration with:
- University research labs working on maritime autonomy
- Defense contractors building USV/AUV systems
- Maritime simulation companies seeking an AI-ready platform
Contact: navisynmarinesolutions@gmail.com
Synergized Navigation Autonomy
NaviSyn Marine Solutions is an early-stage deep-tech startup developing simulation infrastructure for the autonomous maritime industry. Our mission is to accelerate the development of safe, intelligent marine vessels through accessible, research-grade simulation tooling.
© 2026 NaviSyn Marine Solutions · All rights reserved
NaviSense Simulator is proprietary software. Contact us for research licensing.