elevation_mapping_gpu_ros2

This repo contains the implementation of elevation mapping cupy from ETH for ros2 humble, following this PR. Issues and planning related to the elevation mapping. We can share the code, issues, etc.. here

Build Docker Image

In order to build the Docker image with all the required dependencies to run the elevation mapping cupy, run the following command. This will install a Ubuntu 22 environment with ROS2 Humble and CUDA 12.1 for x64 architecture machines.

Clone this repo:

git clone git@github.com:iit-DLSLab/elevation_mapping_gpu_ros2.git

Build the Docker:

sudo docker build -f elevation_mapping_cupy/docker/Dockerfile.x64 -t elevation_mapping_cupy_ros2:latest .

Enter the Docker:

cd elevation_mapping_gpu_ros2
./elevation_mapping_cupy/docker/run.sh

To launch the elevation map:

1. In Terminal 1 (T1):

ros2 run rmw_zenoh_cpp rmw_zenohd

2. Find the container name with docker ps, then in T2:

docker exec -it elevation_mapping bash
ros2 launch elevation_mapping_cupy elevation_mapping.launch.py

If you want to use the code with a rosbag, then you have two options:

1. Option 1: Change the path here, with the path where you rosbags are usually stored
2. Option 2: Run the Docker image with ./elevation_mapping_cupy/docker/run.sh /your_path/to/rosbags

In both cases, when you enter the Docker you see two folders in your Docker home: workspace and rosbags. The first one is where you can access, possibly change (if needed), and build the code of the elevation mapping, where the second one is where you can run your rosbags with ros2 run your_rosbag --loop.

To visualize the grid_map on RViz: - Click Add --> By display type --> GridMap - Select GridMap --> then set the Topic to /elevation_mapping_node/elevation_map_raw - In the GridMap panel --> use the Layer dropdown to pick e.g. "elevation" or "traversability" You can do the same with the /elevation_mapping_node/elevation_map_filter topic