RASA project

Task-oriented dialogue system for talking to a robotic manipulator. Built using the RASA chat-bot framework. This was developed for the Little Helper 7+ at Aalborg University, however it can also be run as a stand-alone application.

Requirements

pip3 install -r requirements.txt --user

Use Guide

The following commands should be run from the rasa_project folder. Note: if you get a command not found error when running the rasa commands below even though rasa is installed, use python3 -m rasa instead.

Model

A model is already present in the repository (file models/model.tar.gaz). It was trained using the pipe-line specified in config.yml using all the training data in the data folder.

To re-train the model:

rasa train --domain=./domain.yml --config=./config.yml --fixed-model-name=model

Running the chat-bot

Launch the action server

Run the actions server (handlers defined in rasa_project/actions.py):

rasa run actions

This will run an action server at http://localhost:5055 Note: make sure the action endpoint is specified in endpoints.yml.

Talk from the command line

To talk to the chatbot in the command line:

rasa shell

To run the chatbot in interactive mode (where the user gives feedback about whether the intent, entities and response are correct):

rasa interactive -m .\models\model.tar.gz

Talk from the web-app

In three separate terminals:

Launch the web-app

cd rasa-webapp && python3 main.py

Start a RASA assistant

rasa run --enable-api -p 5005

This will start a RASA assistant on http://localhost:5005

Start an HTTP server for sending the image files to the client

python -m http.server 8888

Now open http://localhost:8080/ in a browser to see the voice interface.