Rasa Conversational AI

Create conversational experiences with characters, bots and avatars in VR / AR / 3D / 360 across devices and platforms with Rasa and LearnBrite. Have fun!

Jump to:

RASA / LearnBrite Character Conversational Experiences

Before you start (requirements)

Example server setup with AWS

Setup AWS EC2 instance

Installation of Rasa and RasaX

Launch and shutdown of Rasa and RasaX

Install an SSL certificate in order to connect to Rasa using HTTPS

Setting up the 3D environment

Create a Space

Enable Speech

Add Bot to Space

Request activation of the Rasa Gateway for your space

You’re done!

Using Rasa for ChatMapper choices

Showing both speech recognition and text choices

FAQs

I get an error (“CORS error”) in the developer tools when a request is made to my webhook

I want to have speech recognition, but also show clickable choice buttons

RASA / LearnBrite Character Conversational Experiences

Rasa, combined with LearnBrite, can build fully autonomous AI driven avatars. Build a sales bot, advice bot, info bot or companion bot. Connect your Rasa AI agent to a LearnBrite avatar and begin talking to it. Just follow this tutorial. A little Javascript knowledge is handy.

Before you start (requirements)

This tutorial assumes you already understand and are knowledgeable with setting up and building your own RASA chatbot and deploying an instance. RASA is open source software for building online contextual chatbots and AI assistants. Build your own chatbot now. https://rasa.com/docs/rasa/ Need help? Visit https://glasir.ai to get hired help and get started quickly. 

Example server setup with AWS

Setup AWS EC2 instance

NOTE: standard storage is not sufficient – Costs are upwards of US$50 – US$60 per month

# minimum instance type: t3a.large

# minimum storage (to be selected during instance creation): gp2 100 GiB

  1. Login to your AWS Console – https://aws.amazon.com/developer/
  2. Navigate to EC2 Instance and click "Launch Instance"
  3. Select Ubuntu Server 20.04
  4. Select General Purpose t3a.large
  5. Select "Configure Instance Details"
  6. Select "Next: Add Storage"
  7. Select Volume Type: General Purpose SSD (gp2)
    Size (GiB) 100
  8. Click "Next: Add Tags"
  9. Click "Next: Configure Security Group"
  10. Add the following security groups.

  1. Click "Review and Launch"
  2. Click "Launch"
  3. Select an existing key pair or create a new key pair.
  4. Click "Launch Instances" Note: Cost is minimal for 1st week.
  5. Click "View Instances"
  6. Click your "Instance ID"
  7. Click "Connect" Navigate to "SSH Client" Tab and follow connection instructions

Once you have connected to your server via terminal follow the next steps to setup Rasa.

Installation of Rasa and RasaX

Enter following commands into terminal    

1. Connect to SSH Amazon Ubuntu instance

ssh -i "YOUR-AMAZON-KEY.pem" YOUR-AMAZON-INSTANCE

2.

sudo curl -sSL -o install.sh https://storage.googleapis.com/rasa-x-releases/0.31.5/install.sh

3.

sudo bash ./install.sh

Follow any prompts

4.

cd /etc/rasa
sudo docker-compose up -d

5.Install python3 if not installed

sudo apt install python3
sudo python3 rasa_x_commands.py create –update admin me YOUR-PASSWORD

6. At this point both Rasa and RasaX should be active on the AWS instance and the RasaX console should be reachable at http://YOUR-SERVER-IP/login 

This is then the moment to:

  • log in to RasaX using the instance ip address followed by /login
  • set up the link to a Github repository
  • upload a trained rasa model file or train a model using the RasaX menu
  • Use this test model in your own git repository: https://github.com/R1ftwarr/rasatestbot

NOTE 1: the RasaX web interface has some refresh issues, so after connecting a repo it might be necessary to refresh the browser

NOTE 2: in order to enable the REST API for Rasa, log to the instance and edit credentials.yml adding the following line if not already present:

rest:

7. Save the file and restart the containers with:

sudo docker-compose down
sudo docker-compose up -d

Launch and shutdown of Rasa and RasaX

The AWS instance can be stopped and reactivated depending on usage.

In order to stop or launch the Rasa components the AWS instance must be active and login from a terminal is required.

After logging in:

cd /etc/rasa

# Start Rasa and leave control to the terminal (-d = detached)
sudo docker-compose up -d

# Shut down Rasa
sudo docker-compose down

NOTE: Run before creating SSL certificate

Install an SSL certificate in order to connect to Rasa using HTTPS

The AWS instance must be active and login from a terminal is required. After logging in:

sudo apt install certbot
sudo certbot certonly

NOTE: Choose option 1. And spin up a server

sudo ls /etc/letsencrypt/live/<YOURDOMAIN.COM>/privkey.pem
sudo cp /etc/letsencrypt/live/<YOURDOMAIN.COM>/privkey.pem /etc/rasa/certs/
sudo ls /etc/letsencrypt/live/<YOURDOMAIN.COM>/fullchain.pem
sudo cp /etc/letsencrypt/live/<YOURDOMAIN.COM>/fullchain.pem /etc/rasa/certs/

Setting up the 3D environment

Create a Space

Enable Speech

Add Bot to Space

Click the cog/settings icon above the bot. Copy the avatar_externalid and save it . This will later be added to  “YOUR_AVATAR_EXTERNAL_ID” code snippet. Close the window.

Click the conversation icon above the bot

The conversation editor will appear. Click the More button near the bottom, choose the Script action under General.

Paste the following code in the Script field (adjusting to match your avatar_externalid)

var rasaUrl = "YOUR_RASA_ENDPOINT_URL";

var botId = "YOUR_AVATAR_EXTERNAL_ID";

LB.Actions.pauseDialog({executedFromEA:true});

setTimeout(function(){

  LB.uiController.choices.create([{id:0,text:"Cancel"}], "Speak now", LB.uiController.choices.remove);

  LB.cmPlayer.speechRecognition.srStart_listenForCommand(function(data){

    LB.uiController.choices.remove();

    startRasa(data.results[0][0].transcript, botId, rasaUrl);

  });

},100);

Click the Add action button and Save button.

NOTE: Remove any other action in your conversion path except the script action.

Request activation of the Rasa Gateway for your space

Contact LearnBrite support providing the URL to your space requesting for Rasa Gateway to be added to your space. They will add the gateway script to your space,

https://cache.learnbrite.com/_assets_/partners/rasa/rasa.js

 

NOTE: This script assumes that your Rasa Url endpoint will return an reply object array formatted like this:

[{"text": "First text reply"}, {"text":"Second text reply"}]

The text of the last reply from wit.ai is also stored in a variable called Rasa_LastReply. This variable is available across the various access methods, e.g.

// In a script
LB.Actions.displayNotification({
  message: LB.Actions.getUserVariable({ name: "Rasa_LastReply" }),
  type: "warning",
});

// In a ChatMapper script
LB.Actions.displayNotification({
  message: Variable["Rasa_LastReply"],
  type: "warning",
});

// In the text of a ChatMapper DialogNode
I'm a bot and the answer is [var=Rasa_LastReply]

You’re done!

Once support has added the script you can then open your space, click your bot and talk to it. Your bot will respond however you built your Rasa assistant. Have fun!

Using Rasa for ChatMapper choices

Getting input from the user is easy, but the true power of the speech recognition comes from integrating Rasa with ChatMapper’s branching conversations!

Add the following snippet to the parent of your choices:

var rasaUrl = "YOUR_RASA_ENDPOINT_URL";

var botId = "YOUR_AVATAR_EXTERNAL_ID";

LB.Actions.pauseDialog({executedFromEA:true});

setTimeout(function(){

  LB.uiController.choices.create([{id:0,text:"Cancel"}], "Speak now", LB.uiController.choices.remove);

  LB.cmPlayer.speechRecognition.srStart_listenForCommand(function(data){

    startRasa(data.results[0][0].transcript, botId, rasaUrl);

  });

},100);

The answer of the Rasa speech recognition is saved to the variable Rasa_LastReply, which you can use as a condition in ChatMapper, e.g.:

Variable["Rasa_LastReply"] == "pizza"

To use it as a condition, you can therefore organize your nodes as follows:

Showing both speech recognition and text choices

This is possible but requires additional configuration.

In the parent node of your choices (i.e. the node from which the choices branch out), add the following script:

var rasaUrl = "YOUR_RASA_ENDPOINT_URL";

var botId = "YOUR_AVATAR_EXTERNAL_ID";
var choices = [
 {id:CM_NODE_ID, text:"My button text"},
 {id:OTHER_CM_NODE_ID, text:"My other button text"},
 {id:0, text:"Cancel"},
];

LB.Actions.pauseDialog({executedFromEA:true});
setTimeout(function(){
 LB.uiController.choices.create(choices,"Speak now",onChoiceSelected);
 LB.cmPlayer.speechRecognition.srStart_listenForCommand(function(data){
   LB.uiController.choices.remove();
   startRasa(data.results[0][0].transcript, botId, rasaUrl);
 });
},100);

After you pasted the snippet in ChatMapper, you should paste in your Rasa URL and the ID of the bot which will speak the answer (if any) as well as fill in the “choices” variable, which currently looks like this:

var choices = [
 {id:CM_NODE_ID, text:"My button text"},
 {id:OTHER_CM_NODE_ID, text:"My other button text"},
 {id:0, text:"Cancel"},
];

You might have noticed that the choices have a specific format:

{id: <ChatMapper node ID>, text: "<Text that appears on the button>"},

  • Each choice line is wrapped by { }, and ends with a comma
  • The ID is a number (the Cancel button has a special ID, 0)
  • The “text” part is wrapped in quotes

Missing any of these will probably result in the choices not working. You can of course add as many choices as you want, for example the following will show 4 buttons (including Cancel).

var choices = [
 {id:4, text:"Create account"},
 {id:2, text:"Show tutorial"},
 {id:9, text:"Navigate to next floor"},
 {id:0, text:"Cancel"},
];

To make choices progress with speech recognition as well, see the section on Using Rasa for ChatMapper choices.

FAQs

I get an error (“CORS error”) in the developer tools when a request is made to my webhook

This is due to the CORS (“Cross Origin Resource Sharing”) policy set on your server. Amazon automatically provides the correct response headers to enable cross-origin requests, if you are hosting Rasa/RasaX on your own server you must be sure to enable them as well, see: https://developer.mozilla.org/en-US/docs/Web/HTTP/CORS#the_http_response_headers 

How to use speech recognition and clickable choice buttons at the same time

See the “Showing both speech recognition and text choices” section.

How useful was this article?

Click on a star to rate it!

We are sorry that this article was not useful for you!

Let us improve this article!

Tell us how we can improve this article?