Documentation Status

Cannot load information on name: PACKAGE, distro: electric, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.
Cannot load information on name: PACKAGE, distro: fuerte, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.
Cannot load information on name: PACKAGE, distro: groovy, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.
Cannot load information on name: PACKAGE, distro: hydro, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.
Cannot load information on name: PACKAGE, distro: indigo, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.
Cannot load information on name: PACKAGE, distro: jade, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.
Cannot load information on name: PACKAGE, distro: kinetic, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.
Cannot load information on name: PACKAGE, distro: lunar, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.
Cannot load information on name: PACKAGE, distro: melodic, which means that it is not yet in our index. Please see this page for information on how to submit your repository to our index.

Overview

This package uses the Google Text-To-Speech (TTS) API to send results to Dialogflow, Google's NLP platform.

Installation

There is an install.sh script available in git directory if you wish to use it, however, I will go over the steps one-by-one here. Installing this package requires 3 main steps: cloning the dialogflow repo, setting up your Google cloud project, and setting up Dialogflow. However, we need to install PortAudio so we can use PyAudio to get mic data.

sudo apt-get install portaudio19-dev

Cloning The Repo

Install all the requirements using pip by cloning the Github repo and installing all the packages in requirements.txt.

    cd ~/catkin_ws/src
    git clone https://github.com/piraka9011/dialogflow_ros.git
    cd dialogflow_ros
    pip install -r requirements.txt

Google Cloud Setup

Follow the instructions here for configuring your Google Cloud project and installing the SDK for authentication. You will need a google/gmail account.

Usage of the Google Cloud SDK requires authentication. This means you require an API key and an activated service account to utilize the APIs.

  1. Setup a service account

  2. Download the service account key as a JSON.
  3. Check you have GOOGLE_APPLICATION_CREDENTIALS in your environment. This should be the path to the keys.

export GOOGLE_APPLICATION_CREDENTIALS='/path/to/key'
  1. Run the authentication command:

gcloud auth activate-service-account --key-file GOOGLE_APPLICATION_CREDENTIALS

Dialogflow Setup

Follow the steps here to setup authentication with Dialogflow. Note the name of your project-id and make sure to change that in config/params.yaml.

Usage

Follow the steps below to run the package properly.

Configure topics

Go into the config directory and change the following parameters in the params.yaml file:

  1. results_topic: (Optional) The topic where your results will be published.

  2. project_id: The name of your project for the Google Speech node. This is the name of your Google Cloud project when going through the Google Cloud setup.

Launching nodes

To start the Dialogflow nodes, run the following command:

roslaunch dialogflow_ros dialogflow.launch

ROS Nodes

mic_client

ROS node receives text from the Google Cloud Speech API and publishes it onto text_topic (see config/params.yaml). This is used by the dialogflow_client node.

Published Topics

text_topic (std_msgs/String)
  • Acquired text from the Google Cloud Speech API.

dialogflow_client

ROS node that takes text from the mic_client node and sends it to Dialogflow for parsing.

Published Topics

results_topic (dialogflow_msgs/DialogflowResult)
  • Publishes a message with the actions, parameters (python dictionary), and fulfillment text associated with the detected intent as a std_msgs/String.

Troubleshooting

  • If you get an OpenSSL error/InsecurePlatformWarning, run sudo pip install 'requests[security]'. Otherwise make sure you are running Python version 2.7.9+.
  • If you get any ALSA lib pcm.c ... errors, see this thread.

  • Any underrun errors means we need to change chunk size.
  • If you get a bt_audio_service_open error, run sudo apt-get remove bluez-alsa

Wiki: dialogflow_ros (last edited 2018-03-29 10:48:44 by AnasAbouAllaban)