Skip to content

README.md

Latest commit

 

History

History
87 lines (58 loc) · 4.5 KB

README.md

File metadata and controls

87 lines (58 loc) · 4.5 KB
Raw

Multimodal-Wireless Dataset: Extension and Replay Toolkit

This guide provides instructions on how to set up the environment and software needed to extend or replay the Multimodal-Wireless dataset.

Table of Contents

Prerequisites

Before you begin, ensure you have the following tasks done:

  • Clone the current github repository
  • Download the Blender asset from 'Blender_Towns' folder in Dataset or 中国大陆镜像

If you only want to replay some scenario:

Installation Guide

Please follow these steps carefully to configure your system.

1. Setting up the Python Environment

First, we will create a dedicated Conda environment to manage all Python dependencies.

  1. Create and activate the Conda environment: Open your terminal and run the following commands. This will create a new environment named mmw with Python 3.10.

    conda create -n mmw python==3.10
conda activate mmw

  2. Install required packages: Once the environment is active, install all the necessary Python libraries using the requirements.txt file.

    pip install -r requirements.txt

2. Setting up Blender with Mitsuba Renderer

This project relies on Blender and the Mitsuba renderer for visualization and data generation.

  1. Download Blender: Download and install Blender version 4.3.2 from the official website: blender.org/download

  2. Download Mitsuba Add-on: Download the mitsuba-blender.zip file from the official Mitsuba-Blender GitHub releases page.

  3. Install the Add-on in Blender:

    • Open Blender.
    • Navigate to Edit → Preferences → Add-ons.
    • Click the Install... button and select the mitsuba-blender.zip file you just downloaded.
    • Enable the add-on. The installation is successful if you see a 'Mitsuba (.xml)' option under File > Export.

  4. Install PyYAML for Blender: You need to install the pyyaml package into Blender's internal Python environment. You may need to do this from the command line using Blender's Python executable.

    Note: The exact command depends on your operating system and where Blender is installed. Find the Python executable inside your Blender installation folder.

3. Setting up CARLA Simulator

This project is compatible with a specific version of the CARLA simulator.

  1. Download CARLA: Download CARLA version 0.9.15 from the official GitHub release page: CARLA 0.9.15 Releases

How to Extend/Replay

Once your environment and software are fully set up, you can begin the extension or replay process by following the detailed steps in our tutorial video.

Generally, the generation of dataset includes three steps:

  • CARLA: Define the scenario in CARLA via the config file and implement it via carla_main.py, which produces five sensor data and the yaml file of all frames simultaneously.
  • Blender: Reconstruct the whole scenario frame by frame in Blender based on the .yaml file of each frame.
  • Sionna: Load the scene of each frame and compute multi-path components and generate the corresponding channel.

➡️ Watch the Tutorial on YouTube or Bilibili

Note: ./Sionnal_channel_storage/channel_storage_v2i.py collects channels with LOS paths, with the reflection mode on and the scattering mode off. To dump pure NLOS channels, one should refer to ./Sionnal_channel_storage/channel_storage_v2v.py which involves the scattering paths calculation when no LOS paths or other reflection paths exists.