uva-robotics / ros2_pepper

Scripts for cross-compiling ROS and ROS2 for Softbank's Pepper
Apache License 2.0
0 stars 0 forks source link

Introduction

This project contains a set of patches and scripts to compile and run ROS 1 and ROS 2 onboard a Pepper robot, without the need of a tethered computer.

Pre-requirements

Download and extract the NaoQi C++ framework and Softbanks's crosstool chain and point the AL_DIR and ALDE_CTC_CROSS environment variables to their respective paths:

export AL_DIR=/home/NaoQi  <-- Or wherever you installed NaoQi
export ALDE_CTC_CROSS=$AL_DIR/ctc-linux64-atom-2.5.2.74

Prepare cross-compiling environment

We're going to use Docker to set up a container that will compile all the tools for cross-compiling ROS and all of its dependencies. Go to https://https://www.docker.com/community-edition to download it and install it for your Linux distribution.

  1. Clone the project's repository
$ git clone git clone https://github.com/esteve/ros2_pepper.git
$ cd ros2_pepper

ROS 1

Prepare the requirements for ROS 1

The following script will create a Docker image and compile Python interpreters suitable for both the host and the robot.

./prepare_requirements_ros1.sh

Build ROS 1 dependencies

Before we actually build ROS for Pepper, there's a bunch of dependencies we'll need to cross compile which are not available in Softbank's CTC:

./build_ros1_dependencies.sh

Build ROS 1

Finally! Type the following, go grab a coffee and after a while you'll have an entire base ROS distro built for Pepper.

./build_ros1.sh

Copy ROS and its dependencies to the robot

By now you should have the following inside .ros-root in the current directory:

We're going to copy these to the robot, assuming that your robot is connected to your network, type the following:

$ scp -r .ros-root nao@IP_ADDRESS_OF_YOUR_ROBOT:.ros-root

Run ROS 1 from inside Pepper

Now that we have it all in the robot, let's give it a try:

SSH into the robot

$ ssh nao@IP_ADDRESS_OF_YOUR_ROBOT

Source (not run) the setup script

$ source .ros-root/setup_ros1_pepper.bash

Start naoqi_driver, note that NETWORK_INTERFACE may be either wlan0 or eth0, pick the appropriate interface if your robot is connected via wifi or ethernet

$ roslaunch naoqi_driver naoqi_driver.launch nao_ip:=IP_ADDRESS_OF_YOUR_ROBOT \
    roscore_ip:=IP_ADDRESS_OF_YOUR_ROBOT network_interface:=NETWORK_INTERFACE

ROS 2

BEWARE: The ROS 2 port is still experimental and incomplete, simple sensors such as the bumpers work, but the camera driver has not been ported yet.

The following instructions require that you have ROS 1 built for Pepper.

Prepare the requirements for ROS 2

The following script will create a Docker image and compile Python interpreters suitable for both the host and the robot.

./prepare_requirements_ros2.sh

Build ROS 2

Let's now build ROS 2 for Pepper:

./build_ros2.sh

Copy ROS 2 and its dependencies to the robot

Besides the ROS 1 binaries and its dependencies, we'll now a few more directories inside .ros-root in our current directory:

We're going to copy these to the robot, assuming that your robot is connected to your network, type the following:

$ scp -r .ros-root nao@IP_ADDRESS_OF_YOUR_ROBOT:.ros-root

Run ROS 2 from inside Pepper

Now that we have it all in the robot, let's give it a try:

SSH into the robot

$ ssh nao@IP_ADDRESS_OF_YOUR_ROBOT

Source (not run) the setup script

$ source .ros-root/setup_ros2_pepper.bash

ROS 2 does not have a something like roslaunch yet, so you'll have to run naoqi_driver directly:

Start naoqi_driver, note that NETWORK_INTERFACE may be either wlan0 or eth0, pick the appropriate interface if your robot is connected via wifi or ethernet

$ naoqi_driver_node --qi-url=tcp://IP_ADDRESS_OF_YOUR_ROBOT:9559 \
    --roscore_ip=IP_ADDRESS_OF_YOUR_ROBOT --network_interface=NETWORK_INTERFACE \
    --namespace=naoqi_driver

Demos

The folks at the Universidad Rey Juan Carlos and Intelligent Robotics have produced the following video showing a Pepper robot runnning ROS onboard using the code from this repository:

Pepper Navigation

Enjoy!