The idea came when the team was playing video games in my living room and we had to keep moving the tv in order for the active players to have a better viewing angle. That's when we got the idea to create a smart tv wall mount.
The team first met in March 22nd and assigned tasks and derived a plan to work on the project. We wanted to create 5 modes:
Mode 1: Move to position (integer 1-11) --> The TV moves to the specified position
Mode 2: Move (integer 1-11) positions (left/right) --> The TV moves the number of positions to the left or to the right
Mode 3: Come closer (gaming mode, fully extended) --> TV wall mound fully extends
Mode 4: Back to base (fully folded against the wall) --> TV moves back to the fully folded position right up against the wall
Mode 5: Face me (the TV would rotate towards your voice) --> TV finds the direction of the voice and turns towards the source
We immediately started working with Modus Toolbox and looking at the different code examples. We also watched the Picovoice seminar and created our own commands (faceMe, turnToPosition, comeCloser, turnPositionRot, goHome) and our own keywords (Hey TV). We then went on and trained the model and added the keywords and context to our ModusToolbox project.
The main ModusToolbox examples we used were the Picovoice Rhino example and the PWM example.
Later we realized that the most difficult function to implement would be the faceMe. We used the PDM/PCM example and managed to separate the sound signal from the left and right microphone. However, we couldn't find a noticeable time delay, according to which we would find the direction from which the voice was coming from. Therefore we decided to drop this part of the project and focus our time on the rest of the functionality.
Right after that, we understood that we would need a robotic arm that would play the role of the wall mount, and because we wanted to keep the budget as low as possible we went with a wooden 4DOF robotic arm from amazon. After some market research we bought this one: https://www.amazon.de/-/en/gp/product/B086K15F38/ref=ppx_yo_dt_b_asin_title_o08_s00?ie=UTF8&psc=1
Unfortunately, the robot came with no instructions and had to find some online. The video that helped me the most with building the robot is this
This is a timelapse of the building of the robot: https://www.linkedin.com/posts/stathis-zafeiriadis_hackathon-infineon-picovoice-activity-6924615851350388736--0OI?utm_source=linkedin_share&utm_medium=member_desktop_web
After that we made the first voice command tests on the PSOC6.
My first concern had to do with the TV bumping against the wall.
The motors we are using are the servo motors SG90 (turning range 0-180 degrees, 11 positions with 16 degree steps).
Position 1
Position 6
As you can see, position 6 is dead center. For low turning angles +- 30 degrees (positions 4-8) the wall mount doesn't need to extend, but for higher turning angles (positions 1-3 and 9-11) the wall mount needs to extend in order for the TV not to hit the wall.
Position 11
Our solution was to make the wall mount partially extend in the problematic positions.
Here you can find a great video showcasing all the different modes and functionalities, along with the commands and how to use them.
Because the IoT Expanse kit fits on top of the PSOC6, we had to find a way to also use the pins D9-D11 for the PWM control of the Servos.
That's when we used the solder to solder "extra/split" pins to the Kit.
In the end it looked like this (side view of the IoT Expanse Sense Kit and the PSOC6):
This is a simple schematic of our project (IoT Expanse not shown for visibility purposes):
In the attachments you can also find a really nice video we made sort of like an advertisement of our product. Check it out, it's super funny!
Business
We called our product Wall-e for obvious reasons and our tagline is "Don't worry, ask Wall-e"
From a business perspective I really think that our project could be a viable business.
After some very basic research, one can find that smart wall mounts cost upwards of 500 Euros and can only be controlled with an IR remote (so 90s). Adding to that the fact that these kind of wall mounts are already probably using some microcontroller to power and control the dc motors, that means that if the companies switch to the PSCO6 with the IoT Sense Expanse Kit, they would still keep all the capabilities they had, but still expand their products capabilities and offer their customer something much better than what they are currently offering, while keeping the price the same.
The project could either be a standalone product, or even be sold as a conversion kit that would be placed firmly on top of the "dumb" wall mount.
Credits for the music used in the video:
Music by prazkhanal from Pixabay.
Comments