https://github.com/levichen1126/gnss-localizer
ROS/Autoware GNSS localizer — converts NMEA to map-frame poses with full Transverse Mercator projection and GNRMC (yaw) support.
https://github.com/levichen1126/gnss-localizer
autoware gnss localization nmea ros rtk-gps
Last synced: about 1 month ago
JSON representation
ROS/Autoware GNSS localizer — converts NMEA to map-frame poses with full Transverse Mercator projection and GNRMC (yaw) support.
- Host: GitHub
- URL: https://github.com/levichen1126/gnss-localizer
- Owner: LeviChen1126
- License: mit
- Created: 2025-08-31T16:08:45.000Z (10 months ago)
- Default Branch: main
- Last Pushed: 2025-08-31T18:15:45.000Z (10 months ago)
- Last Synced: 2025-08-31T18:25:07.499Z (10 months ago)
- Topics: autoware, gnss, localization, nmea, ros, rtk-gps
- Language: C++
- Homepage:
- Size: 104 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# GNSS_Localizer
[中文版本](./README_zh-TW.md)
> This project generates `/gnss_pose` from GNSS/NMEA data, transforms it into the map frame, and can be integrated with modules such as `ndt_matching`.
---
## Introduction
### 1. `nmea2tfpose`
- Subscribes: `/nmea_sentence` (from serial GNSS receiver)
- Publishes: `/gnss_pose` (`geometry_msgs/PoseStamped`, frame = `map`)
- Key modifications:
- **Supports GNRMC messages** and integrates `yaw_` (heading).
- Supports plane selection (Plane 13 = NTUT Ningxin Park in Taipei).
### 2. `fix2tfpose`
- Subscribes: `/fix` (`sensor_msgs/NavSatFix`)
- Publishes: `/gnss_pose`
- Function: Converts GNSS Fix directly into pose in the map frame.
### 3. `geo_pos_conv`
Converts latitude/longitude (WGS84) → local Cartesian coordinates (x,y,z).
- **`geo_pos_conv_SPIE.cpp`**
- Implements the **complete Transverse Mercator projection** (rigorous mathematical formulas for converting from the Earth ellipsoid to a planar Cartesian coordinate system).
- High accuracy, suitable for large-area use.
- **`geo_pos_conv_mydicoor.cpp`**
- Uses a **linear approximation**:
```cpp
m_x = (m_lat - m_PLato) * 110773.52...
m_y = (m_lon - m_PLo) * 100914.35...
m_z = 0
```
- Suitable for small local areas (e.g., campus/test field), with lower computation cost.
- Limitation: Errors increase with distance from the origin.
**Recommendation**:
- Large-scale, multi-location tests → use `SPIE`
- Small-area quick demos → `mydicoor` is sufficient
### RQT Graph
---
## Development Environment
- **CMake** 3.10+
- **C++17** compiler
- (For ROS1 build) `roscpp`, `sensor_msgs`, `geometry_msgs`, `tf`, `nmea_msgs`
---
## Installation / Usage Examples
> ⚠️ Note: Please rename the corresponding `CMakeLists.*.txt` file to `CMakeLists.txt` before building.
### ROS (catkin)
```bash
mv CMakeLists.catkin.txt CMakeLists.txt # rename
# package.xml must also be placed in the root directory
cd ~/catkin_ws/src
ln -s /path/to/GNSS-Localizer .
cd ..
catkin_make
source devel/setup.bash
rosrun gnss_localizer nmea2tfpose
```
### CMake (Only for Demo)
```bash
mv CMakeLists.cmake.txt CMakeLists.txt # rename
mkdir -p build && cd build
cmake ..
make -j
./gnss_localizer
```
---
## Integration with Autoware
- ### Setup steps
1. Launch the Serial GNSS node that publishes `/nmea_sentence`.
2. Launch the `nmea2tfpose` node with parameter `plane: 13` (Ningxin Park).
3. In `ndt_matching`:
- Set the initial pose (`initial_pos`) to `gnss`.
- Adjust the `fitness_score` threshold to **1.0** (default 500 is not suitable for RTK-GPS precision).
4. Use `rqt_tf_tree` to verify the TF structure.
---
## Project Structure
```bash
GNSS-Localizer/
├── .gitattributes
├── .gitignore
├── LICENSE
├── CMakeLists.catkin.txt
├── CMakeLists.cmake.txt
├── package.xml
├── images
│ └── rqt_graph.png
├── gnss/
│ ├── geo_pos_conv_mydicoor.cpp # Simplified local coordinate conversion
│ └── geo_pos_conv_SPIE.cpp # Accurate projection-based conversion
└── gnss_localizer/
├── fix2tfpose/
│ └── fix2tfpose.cpp # GNSS Fix → TF Pose
└── nmea2tfpose/
└── nmea2tfpose_core_SPIE.cpp # NMEA Sentence → Pose
```
---
## License and Acknowledgment
- [MIT License](./LICENSE)
- [Autoware](https://www.autoware.org/)