ARCHITECTURE

Hardware Architecture

Raspberry Pi
Central Computational Node and Server

• Gets information from the ultrasonic ranging sensors
• Elaborate data and manage the calls of the waste disposal center
• Handles the turn on/off of the stripes
• Respond at the various client through a REST interface
• Connected through Ethernet or WiFi to the web

User Mobile Phone
Mobile Computational Node

• Receives notification
• Interact with the Server
• Interface with the user through an Android application

Analyzer for Plastic/Polymer Identification
Sensing Node

• It distinguishes packaging from other kind of plastic
• The result will be shown on an integrated screen
• In the simulated system it will be implemented in the PSDM app with the use of Google Vision API

Ultrasonic ranging sensors
Sensing Node

• It is located on the internal part of the lid of the bin
• It measures the level of fullness through ultrasonic waves
• It is connected with the central node (server)

Led stripes indicator System
Acting Node

• Led stripes are controlled by the Raspberry Pi

Software Architecture

Mobile Application (Android Application)
A simple interface allows the user to:

• Receive notifications that reminds him to throw the waste at the closing time or when a bin is almost full.
• Decide to throw the waste in an arbitrary moment.
• Know the colour he must follow to reach the right waste container.
• Know the level of waste in his bins and in the common waste containers.
• Interact with the server running on Raspberry.
• Exploit the Google Vision API

Runs on all devices with AndroidOS 5.0 or later.

Google Cloud Vision API
It enables developers to understand the content of an image by encapsulating powerful machine learning models in an easy to use REST API. It quickly classifies images into thousands of categories (e.g., "sailboat", "lion", "Eiffel Tower"), detects individual objects and faces within images, and finds and reads printed words contained within images
Through this API we make our application able to recognize a small number of typical recyclable and unrecyclable objects.

Server (Python Web Application)

• Initialize the information on the trash container with the interaction with a database
• Receive data about the level of trash from all the sensor
• Send a mail to waste disposal center to notify the full (or almost) fill of a waste container.
• When an user is going to throw the trash, it calculates the right waste container and light up the stripe to lead him.
• Interact with the mobile devices and the sensors.

Runs on Raspberry Pi 3.

Network Architecture

The mobile devices will connect to the server via WiFi.
The server running on a Raspberry Pi 3, which has an integrated WiFi module.
The Raspberry will interact with the sensors via cable (in the demo). In a real context the sensors will have a wireless connection to the Raspberry.

Open issues

• We need to find out the availability and the dimensions of the sensor that recognize the plastic.
• We don’t know the price of the sensor that recognize the plastic, so we use the Google Vision API instead. Unfortunately, this service provides a less accurate recognition.
• We have already met some employees and we are going to talk with a couple of activities’ director.
• We will show our project with a scaled model on the LADISPE.