how to make (almost) anything

miana smith

final project tracking page (in progress)

the "vending" machine

This is a vending machine which dispenses items based on image recognition. After paying, a selection is made by making the "correct" facial expression into a camera. The vending machine then dispenses a complimentary product to the expression/emotion.

rendering of final

The image is a very rough rendering I made of the final project in AutoCAD. Note that the ground plane is along the upper right edge. I used AutoCAD because I'm familiar with it, and I didn't start CADing the final project until I knew that I'd gotten into the class, which was kind of late into the game.

I'm switching over to Fusion360, which so far I absolutely love using. I'm re-building the design in that environment from a more functional point of view/work flow (i.e. components to assembly).

I want to keep the machine small, so it will have a 2x2 item grid. Each slot is actuated by a continuous rotation stepper motor wired to whatever the controller is. A webcam and a small display (probably something like this: display) are turned on once a quarter (or similarly sized thing) is inserted into the coin op. Hopefully, the coin op should be able to mechanically discriminate between differently sized coins. The micro-controller then makes a selection based on what the webcam sees.

In terms of image recognition, I think it'll be easiest to start with just RGB color blocks– so if a high percentage of the pixels in the webcam display are, say red, it vends one thing. If it's unable to make a decision within an allotted time frame, it vends an assigned fourth option.

Eventually though, I want the vending decisions to be made based off of facial expressions, but, tbh I don't have a lot of programming experience and I don't know how tenable this is.

updated: september 18th, 2018

I'M PIVOTING

Phototrophic origami house plants

These are basically Braitenberg vehicles, but they don't move as much. Each plant (a flower, a stalk, and some amount of leaves) is actuated using a servo motor, which is hidden where the roots of the plant would be, and SMA wires embedded into the stalk/petals of the structures. The motor rotates the whole plant so that it points at the light source. The SMA wire wilts/unwilts the plant. It finds light using the difference between the readout of two phototransistors.

There are a good amount of projects from previous years that are functionally the same as this project (light-orienting devices).

here from 2017 from 2012 also 2012

It was also suggested to me that I look into different ways of actuating the plant that would seem more organic, compared to using servo. Such as: pneumatic devices? Like inflatable things? which would be pretty cool.

rendering of final
updated: november/october something something 2018

I've been messing around with SMAs and I think I want to abandon that line of actuation

And servos are kinda noisy and Zach made a good point about them being not particularly organic-looking. So I'd like to pivot away from that, too.

I really like the idea of making a small, light-orienting inflatable structure, but I have absolutely no idea how to do that. Ben says that in the wild card week there's a section on soft robotics, which I assume would explain how to do this, but that's kind of late in the game.

The behavior that I wanted the SMAs to exhibit was basically just tensioning a flexible structure, which could be alternatively achieved using a string wrapped around a small motor which tightens to straighten out the structure and relaxes to, um, relax the structure. And instead of the servo, maybe I should consider driving another motor connected to a somewhat assymetric gear that might look like less rigid motion?

And then given time in the semester, I can try to figure out an air-driven version of this.

updated: november 25

I think the soft robotics route is less unreasonable now. I bought some stretchy silicone and have been playing around with some online tutorials, and I think that this is substantially more viable than the SMA route. Pictures to come.

updated: like yesterday lol

I did inflatables for wildcard week and I made the physical mechanism of this project. Electronics-wise, this means that I need to make a relay board between the inflatable's fan and power which toggles on/off based on what a sensor (either a phototransistor, if I stick with the light thing or a capacitive touch sensor, which I think might be more fun) sees.