RPiCopter, unity, and graphics are all halted with the intent of returning to them when interest or usefulness piques again.
So I am building a smart mirror now!
Here's how the Idea started:
I am moving in to a new place, and my friends have been giving me grief for not having any wall decor. So, I decided to make my own. I had the idea of a Iron Man-esque setup with my own Jarvis helping me out and stuff, as any true mechanical engineer does (and should) want. Step 1: Jarvis.
I debated between using Mycroft open AI, Alexa, and Google Assistant as my base, since I'm obviously way out of my depth in creating my own AI. I settled on google assistant because I'm a Google fanboy (Google please hire/fund me), and also because they have the most straightforward API for the Raspberry Pi 3. Took me several tries, but got it to work, plus they have a Python API, so don't need to work too hard to figure out how to use it.
Hardware wise: I am using a raspberry pi 3, ultrasonic sensor, 7'' HDMI LCD display, logitec webcam, a usb sound card, 3.5mm speakers, and a 3.5mm simple desktop mic.
For the prototype mirror, I bought a mirror reflective window tint film on amazon, and a 12'' x 12'' photo frame form Michael's. I removed the glass from the frame, and applied the mirror film, and replaced the glass. then in the backboard of the frame, I cut out two large rectangles: one for my LCD screen so that it would be in direct contact with the glass, and another one for the camera for the same reason. I blacked out the front of the backboard with black construction paper, and temporarily taped the parts in place. I then attached the ultrasonic sensor to the bottom edge of the frame, because unlike the camera, it is not going to work through the semi-transparent reflective film.
the film doesn't really create a mirror finish, and the reflection looks blurred, but it really is good enough for a prototype. the mirror is ideal under certain lighting conditions. and in others, you can clearly see the mounted display and camera, or the display is too bright and looks a different color. The latter situation will almost definitely be fixable in the future with larger and more affordable OLED displays. because the main reason is because of the backlight.
So I am building a smart mirror now!
Here's how the Idea started:
I am moving in to a new place, and my friends have been giving me grief for not having any wall decor. So, I decided to make my own. I had the idea of a Iron Man-esque setup with my own Jarvis helping me out and stuff, as any true mechanical engineer does (and should) want. Step 1: Jarvis.
I debated between using Mycroft open AI, Alexa, and Google Assistant as my base, since I'm obviously way out of my depth in creating my own AI. I settled on google assistant because I'm a Google fanboy (Google please hire/fund me), and also because they have the most straightforward API for the Raspberry Pi 3. Took me several tries, but got it to work, plus they have a Python API, so don't need to work too hard to figure out how to use it.
Hardware wise: I am using a raspberry pi 3, ultrasonic sensor, 7'' HDMI LCD display, logitec webcam, a usb sound card, 3.5mm speakers, and a 3.5mm simple desktop mic.
For the prototype mirror, I bought a mirror reflective window tint film on amazon, and a 12'' x 12'' photo frame form Michael's. I removed the glass from the frame, and applied the mirror film, and replaced the glass. then in the backboard of the frame, I cut out two large rectangles: one for my LCD screen so that it would be in direct contact with the glass, and another one for the camera for the same reason. I blacked out the front of the backboard with black construction paper, and temporarily taped the parts in place. I then attached the ultrasonic sensor to the bottom edge of the frame, because unlike the camera, it is not going to work through the semi-transparent reflective film.
the film doesn't really create a mirror finish, and the reflection looks blurred, but it really is good enough for a prototype. the mirror is ideal under certain lighting conditions. and in others, you can clearly see the mounted display and camera, or the display is too bright and looks a different color. The latter situation will almost definitely be fixable in the future with larger and more affordable OLED displays. because the main reason is because of the backlight.
The backlight is always turned on, so even when the color of the screen is set to black, i.e. nothing is being displayed, it still emits some light. This is annoyingly noticeable when you turn the lights off in the room: because there is more light on the screen side of the mirror than the outside, all light travels out, and basically turns in to a night light. OLED actually turns off pixels, so there is no backlight effect.
As for the product itself, here's what I envision:
The mirror will be completely voice activated. when you walk toward it, it will turn on, and welcome you, allowing you access to a multitude of features. all other times, it will merely be wall decor. It will have the ability to shown you many preset features including: weather, to-do lists, stock trends and prices, web articles, etc..
The current Idea I am running with however, is that it is a daily reflection (hehe... he... he... get it?) tool, designed to help you form helpful habits in the beginning and end of the day. Such habits could include: morning posture correction, morning routine drill for tooth brushing, flossing, water drinking etc., daily audio journal, daily thankfulness/mindfulness activity. Any and all of these functions could be selected via a paired smartphone app, to allow a personalized user experience.
I have been working on this off and on (more off) for a month, and I have the display, a basic UI, google assistant, the ultrasonic feature, and the camera working. it was suggested to me to put the camera in there for many possible future functions, for example: personalized displays from facial recognition (working on that right now), timelapse collage of photos, security camera mode, video calls, emotion recogniton, and stress level detection form photo analysis.
Work in progress, will add pictures soon. possibly could sell something like this if I made many quality improvements. Until then, it'll be just for me. And it'll be really cool. And you all will be jealous. Google please fund me.