arturmiller / pybind11_examples

This repository shows minimal pybind11 examples using catkin or colcon.

pybind11_examples

This repository shows minimal pybind11 using ROS. This project depends on pybind11_catkin. It is a pybind11 wrapper for ROS packages. Despite its name pybind11_catkin can be used in combination with catkin as well as with colcon.

Installation

Install prerequisites

This project depends on pybind11_catkin. You can install it with sudo apt-get install ros-<distro>-pybind11-catkin. Depending on the ROS distribution <distro> is kinetic, melodic or noetic.

Build with catkin tools

1. Create a catkin workspace or use an existing workspace.
2. Clone this repository into the src/ subdirectory of the workspace: git clone https://github.com/arturmiller/pybind11_examples.git.
3. Compile the code catkin build.
4. Source the config file: source path_to_workspace/devel/setup.bash.

Build with colcon

1. Create a colcon workspace or use an existing workspace.
2. Clone this repository into the src/ subdirectory of the workspace: git clone https://github.com/arturmiller/pybind11_examples.git.
3. Compile the code colcon build.
4. Source the config file: source path_to_workspace/install/setup.bash.

Examples

Calling a C++ function

The first example is calling the C++ print function from Python.
You run this example with rosrun function_call hello_world.py. As output you should see hello world!.

Basic classes usage

The second example shows how to instantiate and use C++ classes in Python.

Run this example with rosrun classes pets.py. This should be your output:

<Dog named 'Bello'>
Bello
woof woof!
woof woof!
meow!

Inheritance

This example shows a Python class inheriting from a C++ class. A virtual C++ function is overriden in Python and called from the C++ base class.

Run it with rosrun inheritance dog.py. You should see something similar to this:

<__main__.Dog object at 0x7f90143954d0>
Fluffy dog Molly!
Fluffy dog Charly!

Plugin

The last two examples show, how to call C++ code in Python. This is can be very handy if you want to write a Python plugin system for a C++ library, as is the case for e.g. Tensorflow. There are multiple advantages of this approach. For example you can rerun the plugin without recompiling and running a debugger is much simpler Python.
The following classes are part of the plugin system:

• PluginRunner: The plugins are stored and called in this module.
• BasePlugin: All plugins inherit from BasePlugin.
• CustomPlugin: A custom plugin implemented in Python.
• PluginWrapper: The PluginWrapper calls the Python Plugin.
• PyBasePlugin: This module makes the base plugin accessible in Python.

Run this example with rosrun plugin plugin_runner. The Python plugin is called, with the resulting output hello world!. Now you can change the string of the print statement and rerun it without recompiling the source code.