panther-navigation

A GitHub template for Panther: creating a map using Slam Toolbox and navigation with localization using Nav2

Running the demo

The given example is configured for Velodyne Puck, but any LIDAR publishing Pointcloud2 or LaserScan data will work with some configuration.

Running lidar interface

Before running the navigation demo you need to start the LIDAR interface.

For Velodyne Puck see: https://github.com/husarion/velodyne-docker

💡 NOTE: Remember to provide static transform between LIDAR and robot frame (default: velodyne and base_link see: https://github.com/husarion/velodyne-docker#run-with-the-husarion-panther-robot).

Download this repository

git clone https://github.com/husarion/panther-navigation

Setup environment

cd panther-navigation
source setup_virtual_desktop.sh
export POINTCLOUD2_TOPIC=velodyne_points # change topic name to match your LIDAR pointcloud2 topic
export USE_SIM_TIME=False

Setup navigation parameters

Navigation parameters for mapping and nav2 are stored inside /config directory. You can modify these files to suit your needs. For example, you can change the costmap observation source topic to match your LIDAR.

Map environment

Before using Nav2 for autonomous driving it is necessary to provide a map of the environment. If you already have a map you can skip this step. In terminal run:

docker compose -f compose.ros1_bridge.yaml -f compose.pc2ls.yaml -f compose.mapping.yaml -f compose.vnc.yaml -f compose.rviz.yaml up
# if you are not using 3D LIDAR run:
# docker compose -f compose.ros1_bridge.yaml -f compose.mapping.yaml up

To access the NUC desktop and Rviz2 interface go to 10.15.20.3:8080/vnc_auto.html in your browser. You need to specify the password (default: husarion).

slam_toolbox interface for mapping the environment will be launched. Map will be automatically saved every 5 seconds and stored in the maps directory. You can drive the robot around using the WebUI interface available at http://10.15.20.2:8000 in your browser. After mapping is complete you can terminate docker containers with ctrl+c.

Run navigation with localization

Run navigation with localization.

source setup_virtual_desktop.sh
docker compose -f compose.ros1_bridge.yaml -f compose.pc2ls.yaml -f compose.nav2.yaml -f compose.vnc.yaml -f compose.rviz.yaml up
# if you are not using 3D LIDAR run:
# docker compose -f compose.ros1_bridge.yaml -f compose.nav2.yaml up

To access the NUC desktop and Rviz2 interface go to 10.15.20.3:8080 in your browser. You need to specify the password (default: husarion).

To drive the robot around use Rviz2. Specify the robot goal position by choosing 2D Goal Pose and clicking on the provided map. The robot should generate a valid path and follow it. You can also use the 2D Pose Estimate button to fix the robot's position on the map.

Running in simulation

Simulation

Example demo with Navigation2, using The Husarion Panther robot equipped with Velodyne Puck.

Download this repository

git clone https://github.com/husarion/panther-navigation

Setup environment

xhost +local:docker
export POINTCLOUD2_TOPIC=/velodyne_points # change topic name to match your lidar pointcloud2 topic
export USE_SIM_TIME=True

Map environment using slam_toolbox

Before using Nav2 for autonomous driving it is necessary to provide a map of the environment. If you already have a map you can skip this step.

cd panther-navigation
docker compose -f compose.simulation.yaml  -f compose.pc2ls.yaml -f compose.mapping.yaml -f compose.rviz.yaml up

This will run the slam_toolbox interface for mapping environment. A map will be automatically saved every 5 seconds and stored in the maps directory. You can drive the robot around using the WebUI interface available at http://localhost:8000 in your browser. After mapping is complete you can terminate docker containers with ctrl+c.

Run navigation with localization

Run navigation with localization.

cd panther-navigation
docker compose -f compose.simulation.yaml  -f compose.pc2ls.yaml -f compose.nav2.yaml -f compose.rviz.yaml up

To drive the robot around use Rviz2. First, use the 2D Pose Estimate button to fix the robot's position. Specify the robot goal position by choosing 2D Goal Pose and clicking on the provided map. The robot should generate a valid path and follow it.