🦊 Fox CARLA ADAS Simulation Project

A modular, scenario-driven simulation framework built on CARLA 0.9.16 for demonstrating structured ADAS driving scenarios with multi-modal sensor data and physics-based radar synthesis.

---

🚀 1. Prerequisites

Before setting up the project, ensure you have the following installed:

• Operating System: Windows 10/11 (Required for CARLA 0.9.16 and Shenron synthesis).
• CARLA Simulator: Version 0.9.16 (Must be installed and reachable).
• Miniconda/Anaconda: For environment management.
• NVIDIA GPU: Required for high-fidelity radar synthesis and CARLA rendering.

---

🛠️ 2. Environment Setup

1. Clone the Repository

Clone this repository into your CARLA PythonAPI folder:

cd C:\Path\To\CARLA_0.9.16\PythonAPI
git clone <repository_url> Fox
cd Fox

2. Create Conda Environment

We use Python 3.12 for compatibility with modern libraries and CARLA 0.9.16.

Option A: Using environment.yml (Recommended)

conda env create -f environment.yml
conda activate carla312

Option B: Manual Setup

conda create -n carla312 python=3.12 -y
conda activate carla312
pip install -r requirements.txt

3. Install CARLA Python API

Regardless of the option above, you must install the CARLA 0.9.16 wheel:

pip install ..\carla\dist\carla-0.9.16-cp312-cp312-win_amd64.whl

---

🖥️ 3. Usage & Execution

🖱️ One-Click Dashboard (GUI)

The easiest way to run the orchestrator is via the Flask dashboard:

1. Launch CARLA (e.g., CarlaUE4.exe).
2. Double-click dashboard.bat in the project root.
3. Navigate to http://127.0.0.1:5000 to manage scenarios and view telemetry.

⌨️ Command Line Interface (CLI)

You can run scenarios directly via the run.bat wrapper:

• List Scenarios: .\run.bat --list-scenarios
• Run Scenario: .\run.bat braking --frames 300
• With Parameters: .\run.bat cutin --params "TARGET_SPEED=40,CUTIN_DIST=15"
• Skip Processing: .\run.bat showcase --skip-shenron --skip-mcap

---

🏗️ 4. Project Architecture

• src/: Core logic including PipelineManager, SensorManager, and Recorder.
• scenarios/: Modular scenario implementations (e.g., braking.py, cutin.py).
• scripts/: Data conversion tools (MCAP) and the Shenron radar synthesis engine.
• dashboard/: Flask-based web interface for the orchestrator.
• intel/: Deep-dive documentation, project history (CHRONICLES.md), and versioned changelogs.

---

🛰️ 5. The Data Pipeline

The simulation follows a structured multi-stage pipeline:

1. Simulation Stage: Captures RGB, LiDAR, and raw Radar data from CARLA.
2. Shenron Stage: Synthesizes high-fidelity physics-based radar heatmaps.
3. MCAP Stage: Converts all data into a single .mcap file for Foxglove visualization.
4. Video Stage: Generates .mp4 previews of the session.

---

📂 6. Replication Checklist for New PC

If you are moving this project to another machine:

1. Install CARLA 0.9.16.
2. Verify CUDA drivers are up to date for Shenron synthesis.
3. Create carla312 environment and install requirements.txt.
4. Ensure run.bat and dashboard.bat point to the correct activate.bat path (default: C:\ProgramData\miniconda3\Scripts\activate.bat).
5. Run cmd /c "run.bat showcase --frames 50" to verify the full pipeline end-to-end.

---

🛠️ 7. Troubleshooting & Tips

• GPU Idle Mode: If the simulator feels "frozen," check if Idle Mode is enabled in the Dashboard. This throttles CARLA to ~0 FPS to save GPU for Shenron processing.
• Stop Flag: If the simulation won't start, check for a tmp/stop.flag file and delete it.
• Environment Activation: If run.bat fails to find conda, edit the script to point to your activate.bat location (usually in C:\miniconda3\Scripts or C:\Users\<User>\miniconda3\Scripts).
• Coordinate Systems: Remember that CARLA uses a left-handed coordinate system (LHS), while Foxglove/ROS2 use right-handed (RHS). The MCAP converter handles this conversion automatically.

---

Generated by Antigravity AI | Version 1.1 "Ares"