Skip to content

LemuelHornsby/NaviSense

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

21 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

NaviSense Dev Status

NaviSense Simulator

Synergized Navigation Autonomy

A high-fidelity Unity-based marine vessel simulation platform for autonomy research, reinforcement learning, and COLREGS-compliant navigation development.

Status MVP Unity Python License


What is NaviSense?

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

Architecture Overview

┌─────────────────────────────────────────────┐
│              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

Features

✅ Built & Active

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)

🔄 In Progress

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

📋 Planned

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

Tech Stack

Unity C# Python PyTorch ROS2 Stable Baselines3 Gymnasium Crest Ocean


MVP Roadmap

◉ 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


Getting Started

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.py

Note: Full setup documentation coming in M4. For early access or research collaboration, see contact below.


Use Cases

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

Funding & Collaboration

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


About NaviSyn Marine Solutions

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.

Releases

No releases published

Packages

 
 
 

Contributors