Final Project
Context and Idea
For my final project, I want to build a "used items vending machine", for people to buy and sell used items without having to go through the hustle of coordinating a transaction.
My first idea was to start off with one box, conencted to wifi so that it can be supported by a designated app, where people could leave an item they want to sell and others can get a notification in their app with a picture of the item, the price and some detailing. The notification will draw on an image captured by a camera that will be mounted on the ceiling of the box and ran through AI to get a description and a price tag.
I started making a parametric model of it when I decided it's ugly and I want more display comparments in my final design and abandoned it.
Scaling
In next iterations, and as I was working on it in parallel to an entrepeneurship fellowship with DesignX, I realized I better start off with more than one item on display. The reason for it being I'd like to get more people involved on earlier stages, and also because I'd like to have a more dynamic display. I'd like to have a few items on display at all times, and have the items change every few days.
There were some discussions about the sizes of the compartments, but eventually I decided to go with a relatively compact 2x4 IKEA Kallax proportions, to make sure it is easy to adapt in our housing complex, knowing if it will be too big it will face more resistance in implementation.
App and UX
I'm thinking of a very basic app, where you could scroll through the current inventory and make a purchase through an out-of-app paying system.
I stabilized the orange pill with some toothpicks that I'd have to model out later on.
The scan resulted in a really clean, watertight cloased mesh.
The advanced (payed version) definitions really helped, as well as the object masking option.
Design Details
For the next step I started planning the details for the cabinet's locking mechanism, the hinges for the doors, and the framework for placing the board with all its components on a transparent door.
I had a long brainstorming session with my husband, who is a mechanical engineer, and we came up with a few ideas for design solustions I could 3d print.
Then, I modeled everything in detail. The idea is to start off of a ready-made Ikea Kallax cabinet and make doors for it. The top shelf will be double the size to allow for the placement of bigger items, specifically a microwave. I started by modeling in Fusion, but then ultimately resorted to Rhino because the snapping capabilities and overall accuracy in Fusion were driving me crazy and took me too long to model the simplest things.
This is the Ikea cabinet I'm counting on and it's modeling, I also went to buy it to that I can make the doors for the "big thing" week's assignment. I actually tried to get it used but there were none available that I could find around Cmabridge.
Ikea has great documentation of their products, which makes it easier to "hack" them and plan specifically for them.
The item I wanted was convenitently available at the nearest location.
I modeled it in 3d on Rhino so that I could work on the details.
3D printed hinges will go into the shleves, leaving 2mm tolerance margins around the polycarbonate door.
The fasteners will hold the PCB in place leaving enough space for the leds - I'll connect them using a no. 4 screw with a fastener.
The locking mechanism is made out of a "house" for the servo fastened to the polycarbonate door and a compatible slit that will be fastened to the inner side of the cabinet. The servo will slide into the slit - locking the door
Here are some pictures from the process of fbaricating the 3d printed connectors on scale for all the lockers. You can see more about the integration of the 3d printed connectors with the polycarbonate doors, as well as the design process, on the "something big" week's page where I focus on watejetting the polycarbonate doors and assembling them onto the cabinet with the connectors.
3D printed hinges will go into the shleves, leaving 2mm tolerance margins around the polycarbonate door.
The fasteners will hold the PCB in place leaving enough space for the leds - I'll connect them using a no. 4 screw with a fastener.
The locking mechanism is made out of a "house" for the servo fastened to the polycarbonate door and a compatible slit that will be fastened to the inner side of the cabinet. The servo will slide into the slit - locking the door
System Diagram
Here's a list of the electronic components I'll need for each locker's board:
- Xiao ESP32-S3 with a camera component - not sure I'd get to using the camera for the class's deadline, but would like to keep the option for later
- SG90 Arduino Servo motor - for opening and closing the lockers
- Samsung LM301Z+ Leds in a neutral white color (3500K), to light up the inner space in the locker when taking pictures - optional
- Keypad 3x4 matrix, for inserting the code by the user for opening/closing the locker
- Red and Green through-hole soldering led lights
- on-surface pin connectors (1x7 for the keyboard and 1x3 the servo)
- One 8.2 Ohm resistor (for the big flash led)
- Two small 200 Ohm resistors for the red and green lights
For the integration and fastening the 3d printed connectors to the polycarbonate door, I need some machine and wood screws, Jen helped me and we ordered the machine screws from McMasters. The wood screws I got from Home Depot. For each locker I'll need:
- x4 Stainless Steel M3 14mm Elen head from McMaster, for fastening the hinges to the polycarbonate door
- x5 Stainless Steel M3 30mm Elen head from McMaster, for fastening the PCB and the servo to the polycarbonate door
- x9 Stainless Steel M3 Flange Nuts (3.7mm height) from McMaster, to avoid placing pointed pressure on the polycarbonate sheet
- x2 3mm Wood Screws to fasten the locking mechanism to the inner wall of the locker, I bought philips
After ordering all screws and exporting the 3d parts for print, I started trying out the integration, you can also check the output page to learn more about the design process of the final board.
//add links to previous weeks with the descriptions about what stage they relate to //pic from testing just with the door //pic of printed parts and spray //video of assemblingDoor Knobs
I also had to design the doors knobs, I connected them to the Servo's frame, to avoid more holes in the door. The screws are intentionally long and get into the knob, to add strength against a power tilting them up and down.
Here's how it would look with just the screws.
Then I found some cute animal heads and booleaned out a place for the bolts and the screw to go through:
To make a clean finish for the print, I sanded, sprayed it with a primer, let dry for 24 hrs, and sanded some more with fine sanding paper, before spraying with color.
Here are the other friends for the knob:
After the integration I realized it would be helpful to have a magnetic close of the door, to make sure it stays in place after closing, and doesn't swing slightly open again, so that the Servo will slide into its place easily. I found these adhesive magnets sheets which I hope will work, if I stick them to the door and waterjet a piece of scrap metal to glue to the locking mechanism that touches it when the door closes.
Considering the power supply to the cabinet, I realized I'll have to decide between connecting the PCBs with a USB-A or a USB-C cables to a power adapter. I decided to go with USB-A cables, because their adapters are sginificantly cheaper. Here are the cords and adapters I'll have to buy: