UAV Mapping Project

A clean, modular Python project for UAV mapping simulation:

1. Split a large map into deterministic tiles.
2. Stitch tiles into a global bird's-eye-view (BEV) map.
3. Partition the stitched map into exploration regions with pixel and metric bounding boxes.

The code is ROS-agnostic and designed for future integration with ROS2 message adapters.

Project Structure

uav_mapping_project/
├── main.py
├── config.py
├── splitter/
│   ├── image_splitter.py
│   └── test_splitter.py
├── stitcher/
│   ├── base_stitcher.py
│   ├── global_map.py
│   ├── grid_stitcher.py
│   └── test_stitcher.py
├── partition/
│   ├── base_partition.py
│   ├── region.py
│   ├── grid_partition.py
│   └── test_partition.py
├── utils/
│   ├── io_utils.py
│   └── visualization.py
└── data/

Requirements

• Python 3.10+
• numpy
• opencv-python
• matplotlib

Install dependencies:

python -m pip install numpy opencv-python matplotlib

How to Run

Run commands from repository root (BEV_Stitcher).

1) Generate synthetic source map and split into tiles

python -m uav_mapping_project.splitter.test_splitter

Outputs:

• uav_mapping_project/data/source_map.png
• uav_mapping_project/data/split_grid_preview.png
• uav_mapping_project/data/tiles/tile_x{col}_y{row}.png

2) Stitch tiles into global map

python -m uav_mapping_project.stitcher.test_stitcher

Outputs:

• uav_mapping_project/data/stitched_map.png

Console also prints map resolution (meters/pixel) and origin.

3) Partition stitched map into regions

python -m uav_mapping_project.partition.test_partition

Outputs:

• uav_mapping_project/data/partition_overlay.png

Console prints each region in:

• pixel coordinates: bbox_pixel
• metric coordinates: bbox_meter

4) Run full end-to-end pipeline

python -m uav_mapping_project.main

This executes split → stitch → partition in sequence and writes results under uav_mapping_project/data/.

Configuration

Edit uav_mapping_project/config.py to customize:

• tile size and overlap
• synthetic map size
• number of partition regions
• camera/UAV parameters (altitude, fx, fy, cx, cy, image size)
• default map origin and fallback resolution

Notes for Integration

• GlobalMap includes:
  • image
  • resolution (meters per pixel)
  • origin (world coordinate for top-left)
• Partition output Region includes both bbox_pixel and bbox_meter.
• No ROS imports are used; conversion to ROS2 messages can be added in a separate adapter layer.