ROS 2 Nav 2 Python Interface (ROS 2 Nav 2 PYIF)

Python facilitates rapid prototyping and offers a wide array of libraries that are utilized in motion planning research. A powerful tool for testing, benchmarking and deploying motion planners,the Nav 2, is however implemented in C++ and does not support python-based motion planners.

Porting Python code to C++ is often complicated and separating the planner into a different application is cumbersome.

The Nav 2 PYIF provides tools for non-invasive integration of Python-based motion planners as native Nav 2 C++ plugins.

See more info.

Requirements

• Ubuntu 22.04.1 LTS
• ROS Humble Distribution
• Nav 2

Setup

cd colcon_ws/src
git clone <todo>
source /opt/ros/humble/setup.bash
cd ..
colcon build

APF Python Example

The APF example demonstrates the functionality of the Nav 2 PYIF. The example includes a params.yaml config file to run the example.

Setup

pip install -e src/python_controller/

Running the example

source install/setup.bash
export TURTLEBOT3_MODEL=waffle
export GAZEBO_MODEL_PATH=$GAZEBO_MODEL_PATH:/opt/ros/humble/share/turtlebot3_gazebo/models
ros2 launch nav2_bringup tb3_simulation_launch.py headless:=False params_file:=<path_to_nav2_pyif>/src/params.yaml 

Creating your own custom Python controller

Create a Python package

source /opt/ros/humble/setup.bash
cd nav2_pyif/src
ros2 pkg create --build-type ament_python <package_name>

Add your <python_controller>.py into nav2_pyif/src/<package_name>/<package_name>/```

Add mandatory overrides for controller plugin in <python_controller>.py

from geometry_msgs.msg import TwistStamped

def <compute_velocity_commands_override>(occupancy_grid, pose, twist):
    cmd_vel = TwistStamped()
    # integration with your own code here
    return cmd_vel

def <set_plan_override>(global_plan):
    # integration with your own code here
    return

def <set_speed_limit_override>(speed_limit, is_percentage):
    # integration with your own code here
    return

Build the package (NB! changes is <python_controller>.py take effect only after building)

Go back into the root folder.
cd ..
colcon build

Install the local package

pip install -e src/<package_name>/

Add the package to src/params.yaml

python_module: "<package_name>.<python_controller>" # NB! ".py" is not added
python_delegates:
  set_plan: "<set_plan_override>"
  set_speed_limit: "<set_speed_limit_override>"
  compute_velocity_commands: "<compute_velocity_commands_override>"