Agent-Oriented Visual Programming for the Web of Things

Overview

This demo was presented at the Dagstuhl Seminar 23081 Agents on The Web.

It showcases a Web-based Agent IDE that allows domain experts to develop JaCaMo agents using a visual block language that was developed with the University of St. Gallen in the context of the IntellIoT European Project.

Smart-farming use case

Once run, the demo shows two autonomous tractors that coordinate to reach the goal of maintaining a grid of fields well irrigated.

The grid is composed of 4x4 fields that can either be, depending on their humidity level:

• Dry when they are beige
• Normal when they are light brown
• Irrigated when they are dark brown

Over time, the humidity level will decay randomly.

The home position for the tractors is (0,0) in the bottom left corner.

The system is composed of two agents that pilot two tractors:

• One (the red triangle) is a soilchecker and is equipped with a sensor that can measure the humidity level of the soil in a field. The tractor is surveying the whole grid periodically in order to measure the humidity in the fields. If the humidity in a field is lower than a set treshold the tractor will communicate the waypoint to the irrigator.

• The other (the green triangle) is an irrigator and is equipped with a water tank that can hold the amount of water to irrigate 4 fields. The irrigator will then idle at the home position (0,0) where it can also refill its tank and as soon as it receives 4 waypoints it will start to irrigate from the closest to the farthest. If the soilchecker finish surveying the fields and is back at the home position the irrigator will continue irrigating until it has reached all the waypoints even if it needs to take a trip without consuming its full capacity.

Components

This demo is based upon three software components:

• Web-based Agent IDE This component is the one that allows to create agents using a visual block languages that allows to define agents with the jason syntax. The visual language is designed to be easier for people to understand, since the target users of the tool are domain expert with little or no experience in programming.
• MAS REST Runtime this component is an implementation of a server wrapping a JaCaMo application. It uses JaCaMo-REST internally to run the multi-agent system that the user wants to execute.
• Web-of-Things Web Simulator This is a simple JavaScript "simulator" for WoT devices. It is a Node.js server that exposes Thing Descriptions and implement the things APIs running simulated models of the devices with a fixed tick.

How to run the demo

Dependencies

To run this demo you will need to have on your machine:

• Node.js with nodemon globally available
• Java 14
• A MongoDB client (e.g. MongoDB Compass)
• optionally Docker or MongoDB installed locally

1) Clone the repository

First of all clone the repository

git clone https://github.com/samubura/dagstuhl-demo-wot-autonomous-tractors.git

Then enter the main folder and run the script

./init_submodules.sh

This will clone the repositories imported as submodules and bring them to the exact commits that were used at the time of the demo (to avoid future updates to break this demo).

2) Run the Web-based IDE

Using Docker

The easiest way to run the IDE is using Docker.

Enter the intelliot-hypermas-explorer folder and add a file named .env with this content:

CLIENT_PORT=8090
SERVER_PORT=8000

MONGO_URL=mongodb://mongo:27017/masDB

Run the script to launch the whole app:

./launchDocker.sh

Standalone

MongoDB is used as the persistent storage to run the application as standalone MongoDB must be installed in your machine and the mongod command can be used in a shell.

Create a folder where the databases will be stored end edit the launchMongoDB.sh as needed. The script to launch the IDE is configured to create multiple databases to have multiple "workspaces".

The default for Windows users will be to create a C:\data folder.

Setup the repository with the script that installs npm dependencies:

./setup.sh

Run the script to launch the whole app with the name of the "workspace" hence the name of the folder that will contain the actual database:

./launchStandalone.sh tractors

3) Populate the Database

Using a MongoDB client, connect to the databse hosted at:

mongodb://localhost:27017 

Use the .json files in the mongoData folder of this repository to import the data in the corresponding collections.

3) Run the MAS REST Runtime

Open a new shell and enter the mas-rest-runtime folder.

Run with Gradle:

./gradlew run

4) Run the WoT Web Simulator

Open a new shell and enter the wot-web-simulator folder

First download the dependencies with:

npm install

Then you can run the application with

./run.sh

5) Use the application

Go to http://localhost:3000 to open the WoT Web Simulator and see the simulation.

Go to http://localhost:8090 to open the Web IDE.

There you can load the code saved on the DB and have a taste of the visual block language. Click on Load and select the agent you want to load from the dropdown.

You can also try out the block language to assemble other agents.

A runtime configuration is already available to run the tractors example. Simply click on Run and select the runtime configuration you want to execute from the dropdown. This will start the multi-agent system on the MAS REST Runtime.

You can follow the MAS logs on the MAS REST Runtime shell.

To stop the MAS click to Inspect Runtime and click on the stop button.