1.AI sorter
Summary Slide¶
Video Clip¶
What does it do?¶
This machine can automatically sort various things by type.
Sorter¶
Move the identified things to the location of the target receptacle and deposits the item into the open receptacle.
Box¶
The Iris Mechanism opening of the receptacle opens automatically (triggered by ultrasonic sensor) as the sorter tray approaches.
Image Recognition System¶
Image recognition by Rasberry Pi camera and Teachable Machine.
By loading a model that identifies things, it is possible to recognize them by type.
Who’s done what beforehand?¶
A self-made sorting system that artificial intelligence automatically performs separation of such troublesome waste
Three boxes are lining up, which is different from the separation of garbage in Japan, but from the left”Recycling garbage” “Landfill waste” “Compost garbage”It is said that it is a sort of separation.
Image recognition and stepping motors are controlled using Raspbery pi3
An AI-Powered Waste Sorting Robot from BHS powered by Sadako’s AI
they had training their deep learning algorithms. The sorts of things they’re solving for are particular objects like polyethylene terephthalate (PET) bottles along with other thermoset plastics that can be recycled. The bottles can either be full of air or crinkled, and also partially obscured by other bits of rubbish. The algorithm is able to handle all these cases in almost real-time with a very high degree of accuracy thanks to the powers of deep learning:
What did you design?¶
I created three major parts
Box, Sorter and Image Recognition System.
What materials and components were used? / Where did they come from? / How much did they cost?¶
3D Printing Parts¶
-
Sorter
-
Box
2D Laser Cutting Parts¶
-
Sorter
-
Box
-
Image Recognition
Kicad¶
-
Sorter
-
final_project_sorter.kicad_sch[.kicad_sch]
-
final_project_sorter.kicad_pcb[.kicad_pcb]
-
-
Box
-
trash_box.kicad_sch[.kicad_sch]
-
trash_box.kicad_pcb[kicad_pcb]
-
Mods¶
-
Sorter
-
Box
Code¶
-
Sorter
- final_project_sorter.ino [.ino]
-
Box
- final_project_trash_box.ino[.ino]
-
Image Recognition
- image_recognition.py[.py]
materials and components¶
Total cost | Box | Sorter | Image Recognition System |
---|---|---|---|
¥54438 (JPY) | ¥5517 (JPY) x 3 = ¥16551 (JPY) | ¥22245 (JPY) | ¥15642 (JPY) |
Box¶
Component | vender | Cost | Required number | Total Cost | Link |
---|---|---|---|---|---|
ATTiny3216 | digi-key | ¥191 (JPY) | 1x3 | ¥573 (JPY) | Link |
Resistor 0Ω | RS components | ¥5 (JPY) | 1x3 | ¥15 (JPY) | Link |
Capacitor 10μF | RS components | ¥36 (JPY) | 1x3 | ¥108 (JPY) | Link |
Pin Header (PSS-410256-00) | vender | ¥299 (JPY) | 1x3 | ¥897 (JPY) | Link |
Servo Motor (HS-311) | akizukidenshi | ¥1,622 (JPY) | 1x3 | ¥4866 (JPY) | Link |
Ultrasonic Sensor (HC-SR04) | vender | ¥640 (JPY) | 1x3 | ¥1920 (JPY) | Link |
Battery 1.5V (エボルタ乾電池 単3形) 12 sets | monotaro | ¥116 (JPY) | 3x4 | ¥1390 (JPY) | Link |
Battery Case | vender | ¥100 (JPY) | 1x3 | ¥300 (JPY) | Link |
Neodymium magnet | daisonet | ¥100 (JPY) | 1x3 | ¥300 (JPY) | Link |
MDF 600mm x 30mm x 4mm | rakuten | ¥240 (JPY) | 1x3 | ¥240 (JPY) | Link |
MDF 600mm x 30mm x 5.5mm | rakuten | ¥242 (JPY) | 2x3 | ¥1452 (JPY) | Link |
Prusament PETG (Prusa Orange) | amazon | ¥4,490 (JPY) | 1 | ¥4,490 (JPY) | Link |
Sorter¶
Component | vender | Cost | Required number | Total Cost | Link |
---|---|---|---|---|---|
ESP32 | Amazon | ¥712 (JPY) | 1 | ¥712 (JPY) | Link |
Resistor 10kΩ | monotaro | ¥16 (JPY) | 1 | ¥16 (JPY) | Link |
TACT Switch | Digi key | ¥72 (JPY) | 1 | ¥72 (JPY) | Link |
Capacitor 10μF | Amazon | ¥26 (JPY) | 1 | ¥26 (JPY) | Link |
Regulator 3.3V 1A SOT223-3 | RS Components | ¥71 (JPY) | 1 | ¥71 (JPY) | Link |
LED RED | vender | ¥55 (JPY) | 1 | ¥55 (JPY) | Link |
Resitor 100Ω | monotaro | ¥31 (JPY) | 1 | ¥31 (JPY) | Link |
Slide Switch | Digikey | ¥77 (JPY) | 1 | ¥77 (JPY) | Link |
Capacitor 0.1μF | Digikey | ¥14 (JPY) | 1 | ¥14 (JPY) | Link |
Stepper Motor Nema17 | RS Components | ¥5220 (JPY) | 2 | ¥10440 (JPY) | Link |
Motor Driver Drv8825 | AliExpress | ¥175 (JPY) | 1 | ¥175 (JPY) | Link |
Servo Motor SG90 | akizukidenshi | ¥440 (JPY) | 1 | ¥440 (JPY) | Link |
AC Adaptor 12V 1A | monotaro | ¥899 (JPY) | 1 | ¥899 (JPY) | Link |
shaft φ8 mm x 800 mm | monotaro | ¥2,013 (JPY) | 2 | ¥4,026 (JPY) | Link |
timing belt and timing pulley set | amazon | ¥1,199 (JPY) | 1 | ¥1,199 (JPY) | Link |
Linear Bushings LM-8UU | misumi | ¥404 (JPY) | 2 | ¥808 (JPY) | Link |
Half Screw with Hexagon Hole | monotaro | ¥549 (JPY) | 1 | ¥549 (JPY) | Link |
Coil type wiring band (コイル型配線バンド) | komeri | ¥228 (JPY) | 1 | ¥228 (JPY) | Link |
MDF 600mm x 30mm x 2.5mm | vender | ¥220 (JPY) | 1 | ¥220 (JPY) | Link |
MDF 600mm x 30mm x 5.5mm | rakuten | ¥242 (JPY) | 2x3 | ¥1452 (JPY) | Link |
Acrylic 2.5mm | Rakuten | ¥550 (JPY) | 1 | ¥550 (JPY) | Link |
screw M3 x 20mm | Rakuten | ¥185 (JPY) | 1 | ¥185 (JPY) | Link |
Image Recognition System¶
Component | vender | Cost | Required number | Total Cost | Link |
---|---|---|---|---|---|
RaspberryPi 4 8GB Model B | vender | ¥11,440 (JPY) | 1 | ¥11,440 (JPY) | Link |
Raspberry Pi Camera | rakuten | ¥898 (JPY) | 1 | ¥898 (JPY) | Link |
Tact Switch | monotaro | ¥71 (JPY) | 1 | ¥71 (JPY) | Link |
Resistor 10kΩ | vender | ¥5 (JPY) | 1 | ¥5 (JPY) | Link |
Power Supply AC Adaptor | amazon | ¥980 (JPY) | 1 | ¥980 (JPY) | Link |
Packaging¶
Component | vender | Cost | Required number | Total Cost | Link |
---|---|---|---|---|---|
Coil type wiring band (コイル型配線バンド) | komeri | ¥228 (JPY) | 1 | ¥228 (JPY) | Link |
Wiring Tube (配線チューブ) | komeri | ¥508 (JPY) | 1 | ¥508 (JPY) | Link |
Shrink Tubing (収縮チューブ) | komeri | ¥118 (JPY) | 1 | ¥118 (JPY) | Link |
Battery case (電池ケースUM4x3) | komeri | ¥308 (JPY) | 1 | ¥308 (JPY) | Link |
Bundling tape (結束テープ) | komeri | ¥428 (JPY) | 1 | ¥428 (JPY) | Link |
Oil-based black spray (ラッカースプレー) | komeri | ¥658 (JPY) | 1 | ¥658 (JPY) | Link |
What parts and systems were made? / What processes were used?¶
-
Input Device¶
-
Box
- Ultrasonic Sensor (HC-SR04)
-
Image Recognition System
- Raspberry Pi Camera
-
-
Output Device¶
-
Box
- Servo Motor (HS-311)
-
Trash Separator
- Stepper Motor (Nema17)
- Servo Motor (SG90)
-
-
Subtractive¶
- Box
- Box body (Laser Cutter Fabrication)
- Box
-
Additive¶
-
Box
- Box Iris Blade (3D Printer Fabrication)
-
Trash Separator
- Trash Separator Body (3D Printer Fabrication)
-
Image Recognition System
- Raspberry Pi Camera Stand (3D Printer Fabrication)
-
What questions were answered?¶
- In performing a series of flows, does the machine ensure idempotency?
Ran the machine multiple times with the same results. (For example, trash that can be turned into charcoal is always taken to the charcoal-able box.)
How was it evaluated?¶
-
Check to see if the image recognition system allows the sorting machine to move and dump the things into the target Box.
The series of flows worked well.
-
Check to see Whether the machine is able to classify more than 3 types of things
Three types of things could be recognized. Trash separation and color-coding of plastic bottle caps.
What are the implications?¶
This project was initially started to solve a problem in Kuriyama-cho, which is that it is difficult for humans to separate garbage. This is because the garbage separation rules in Kuriyama-cho are very strict. As the project progressed, I found that in addition to garbage sorting, I could also color-code the plastic bottle caps used in the Precious Plastic Machine.
I would like to simplify the identification and sorting process in Kuriyama-cho with automatic sorting by AI.
License¶
I have decided to grant the following license
Because I want many people to be involved in this project, I have only granted the Attribution license
I also need to grant a license for the Drag Chain I use to hide the wiring of the stepper motor in my project, so I am granting an equivalent license.
Special Thanks¶
FabLab Kamakura Instructors
- Youka Kawahara
- Osamu Yamamoto
- Yosuke Tsuchiya
- Rico
- Kai Naito
- Jun Kawahara
- Asako Okazaki
FabLab Kuriyama
- Takamitsu Miki
- Oka Yuki
- Toshiki Tsuchiyama
Global Team
- Saverio Silli
- Ted Hung
- Miriam Choi