Synthetic Data Generation

Contains the code for generating data required for the PredictingLocalWindFields in 3D project.

Generates the following data:

• 2D Cityscapes
• Drone positions
• 2D LiDAR scan
• 2D Windflow data
• 3D Wind flow field using Phiflow
• 3D Cityscape using quadtrees for intersection (csv and mat files)
• PointCloud Data from a simulated LiDAR on a UAV using the MATLAB UAB ToolBox

Usage

Install the required packages using the following command:

pip install -r requirements.txt

Run the following command to generate the cityscapes:

python main.py gen cityscapes

Run the following command to generate drone positions:

python main.py gen drone

Run the following command to generate wind flow fields:

python main.py gen windflow

Run the following command to save visualisations of the wind flow fields:

python main.py viz wind --index 10 --export-all

Run the following command to generate 2D LiDAR scans:

python main.py gen lidar

Run the following command to generate demo viz cityscapes:

python main.py viz wind --index 10 --export-all --export-dir data/demoviz --map_size 140

Run the following command to generate transparent cityscapes:

python main.py viz wind --index 10 --export-all-transparent --map_size 140 --export-dir data/transparent

To the run the interactive demo 🎮

python main.py demo --index 0

The MATLAB files present in this repo must be run from the MATLAB editor application.

Notes

The 3D LiDAR simulation is conducted using the MATLAB UAV toolbox. MATLAB requires 4 corners of a rect to create a mesh. The CSV file contains the start and end coordinates for a rect. The .mat files contain the 4 corners of the rect.

Example PointCloud from Cityscape

Example Wind Flow Field with LiDAR Scans

Output File Structure

data/
├─ cityscapes/
│  ├─ city_0.csv
│  ├─ city_1.csv
│  ├─ city_(...).csv
│  ├─ city_60.csv
├─ drone_positions/
│  ├─ city_0.csv
│  ├─ city_1.csv
│  ├─ city_(...).csv
│  ├─ city_60.csv
├─ exportviz/
│  ├─ city_0.png
│  ├─ city_1.png
│  ├─ city_(...).png
│  ├─ city_60.png
├─ matlab_meshes/
│  ├─ city_0.mat
│  ├─ city_1.mat
│  ├─ city_(...).mat
│  ├─ city_60.mat
├─ lidar/
│  ├─ city_0_pos0.npy
│  ├─ city_0_pos1.npy
│  ├─ city_0_pos(...).npy
│  ├─ city_1_pos0.npy
│  ├─ city_1_pos1.npy
│  ├─ city_1_pos(...).npy
│  ├─ city_(...)_pos(...).npy
├─ windflow/
│  ├─ city_0.npy
│  ├─ city_1.npy
│  ├─ city_(...).npy
│  ├─ city_60.npy
├─ pointclouds/
│  ├─ city_0/
│  │  ├─ pointcloud_1.csv
│  │  ├─ pointcloud_(...).csv
│  │  ├─ pointcloud_10.csv
│  ├─ city_1/
│  │  ├─ pointcloud_1.csv
│  │  ├─ pointcloud_(...).csv
│  │  ├─ pointcloud_10.csv
│  ├─ city_60/
│  │  ├─ pointcloud_1.csv
│  │  ├─ pointcloud_(...).csv
│  │  ├─ pointcloud_10.csv

MATLAB Toolbox requirements

• UAV Toolbox
• Aerospace Blockset
• Aerospace Toolbox
• Control System Toolbox
• Navigation Toolbox
• Stateflow
• Simulink
• Simulink 3D Animation