{"id":4256,"date":"2020-04-22T22:32:36","date_gmt":"2020-04-23T02:32:36","guid":{"rendered":"https:\/\/www.hajim.wdev.rochester.edu\/senior-design-day\/ai-sorting-system\/"},"modified":"2020-05-04T14:54:49","modified_gmt":"2020-05-04T18:54:49","slug":"ai-sorting-system","status":"publish","type":"post","link":"https:\/\/www.hajim.rochester.edu\/senior-design-day\/ai-sorting-system\/","title":{"rendered":"AI Sorting System"},"content":{"rendered":"\n

Team Members<\/h2>\n\n\n\n

Tania Cano
Jordan Floyd
Donghwa Han
Celine Wang<\/p>\n\n\n\n

Supervisors<\/h2>\n\n\n\n

Jack Mottley, PhD, Electrical and Computer Engineering, UR
Daniel Phinney, Adjunct Professor, Electrical and Computer Engineering, UR<\/p>\n\n\n\n

Customers<\/h2>\n\n\n\n

We worked with L3Harris while we were still on campus.<\/p>\n\n\n\n

Description<\/h2>\n\n\n\n

The AI Sorting System helps users sort their recyclable items into the correct category (plastic, metal, glass, or other). It takes a photo of a recyclable object, weighs it, and scans it with a metal detector. It inputs this data into a program that classifies the item as plastic, metal, or glass. Another program uses this data to make a final decision (whether the item is plastic, metal, glass, or other). The system includes 4 bins, one for each material. LEDs will light up on one of the bins and its lid will flip open, indicating that the recyclable should be placed in that bin.<\/p>\n\n\n\n

\"Prototype
When an item is classified, the LEDs under the corresponding bin light up and the lid of that bin flips open. (This picture was taken before the “other” bin was added.)<\/figcaption><\/figure><\/div>\n\n\n\n
\"Image
Sample image classification (before the “metal” category was trained). The trained model is about 98.8% sure that the image on the left is glass.<\/figcaption><\/figure><\/div>\n\n\n\n
\"\"
Completed prototype of bins for the four categories (other, plastic, metal, and glass). To the left of the bins is the scale, which is where the items will be placed.<\/figcaption><\/figure><\/div>\n\n\n\n

Parts List<\/h2>\n\n\n\n

PARTS INVOLVING RASPBERRY PI:
<\/strong>Raspberry Pi 4
USB-C Power Supply for Raspberry Pi 4
MicroSD Card (64 GB)
Micro HDMI to HDMI cable (for Raspberry Pi)
IR Sensor
Raspberry Pi Camera<\/p>\n\n\n\n

PARTS INVOLVING ARDUINO:
<\/strong>Arduino Uno
Arduino Power Supply
Arduino Kit
Scale (Arduino Compatible)
LED Strip (Arduino Compatible)
Servos & Servo Arms (4)
Metal Detector
Bins (4)<\/p>\n\n\n\n

OTHER PARTS:
<\/strong>Laptop<\/p>\n\n\n\n

Steps of Implementation<\/h2>\n\n\n\n

Step 1: Detect the Object<\/h3>\n\n\n\n

The diagram below shows connections between the IR obstacle sensor, Arduino, Raspberry Pi, and the Camera.<\/p>\n\n\n\n

<\/p>\n\n\n\n

The IR sensor on the Raspberry Pi senses if an item has come to close proximity. The Raspberry Pi increments the item ID integer (which is saved to a text file).<\/p>\n\n\n\n

Step 2: Take a Picture<\/h3>\n\n\n\n

The Raspberry Pi takes a picture using the camera module. The size of the image is 512 x 384 pixels.<\/p>\n\n\n\n

Step 3: Take the Metal Detector & Scale Readings<\/h3>\n\n\n\n

The Raspberry Pi sends a message to the Arduino by using SPI communication. In response, the Arduino takes the metal detector reading and scale reading. <\/p>\n\n\n\n

SPI communication was chosen because it is synchronous, it allows the master (the Raspberry Pi) to initiate and control all communications, and it is full-duplex (data can flow in both directions simultaneously).<\/p>\n\n\n\n

Two wires were used to connect between the microcomputer of the metal detector and the Arduino. The following video shows the Arduino reading the signal from the metal detector:<\/p>\n\n\n\n