Automatic Garage Door Opener With Xilinx Kria KV260

The Idea

The humble garage door opener remote. Usually taking the form of a palm-sized box with a few buttons, these haven't changed much functionally since their original invention. And why should they? They work just fine. They're simple to manufacture, easy to use, and handily clip to the sun visor or disappear into the glovebox of a car.

So why, then does Tesla sell a $325 "Automatic Garage Door Opener"? And why are people buying it? Simple: because sometimes it's nice to try something new, especially if it promises additional convenience. However not all of us own Teslas. And not all of us want to install additional hardware in our car when the aforementioned remote control works just fine.

Thankfully a simpler alternative exists in the form of a computer vision application running on a Xilinx Kria KV260 SOM computer installed in one's house. Drive up to your house and the system automatically recognizes and authenticates your vehicle and triggers the garage door to open.

Project Goals

Given the constraints on my living situation, some of these may not be possible to complete. However, even just the computer-vision components related to identification and authentication will be enough to prove that such a system could work reliably and the final triggering of the garage door is just the cherry on top.

• Identify my vehicle on approach to the garage door.
• Read the license plate on my vehicle, if camera resolution allows.
• Recognize an authentication action such as a wave or the flash of headlights.
• Publish an event to an MQTT broker from the KV260.
• Read the event on a NodeMCU located in the garage and trigger a connected relay, causing the garage door opener to activate.

Project Work Timeline

• 12/2/2021: Received Kria KV260 Starter Kit in the mail.
• 1/9/2022: First boot of KV260. Unable to proceed further due to absence of onboard WiFi module or availability of ethernet network on-site.
• 1/14/2022: Modded a Creality WiFi Box to serve as a WiFi-to-Ethernet bridge adapter for KV260, attempted to update the linux system. Stability of connection caused repeated failures to download packages. Ordered a USB WiFi adapter.
• 1/18/2022: Received adapter, attempted to install relevant Mediatek drivers, was unsuccessful.
• 3/1/2022: Ordered a Realtek-based WiFi USB adapter.
• 3/5/2022: Received adapter, attempted to install relevant Realtek drivers, was unsuccessful.
• 3/6/2022: Was able to successfully boot KV260 headless and SSH into it via the available ethernet network in the new location. Attempted to setup VNC server and connect to it but was only able to reach a blank gray screen.
• 3/26/2022: Attempted to utilize [hdcoe]'s guide to setup X2Go as an alternative to VNC, was unable to complete due to client computer limitations.
• 3/27/2022: Published raw project notes for future reference to GitHub.

Project Status

TL;DR: Owing to complications in my own living situation at the moment, this project is still very much so a work in progress and is not complete. Work on it will resume when it can.

The events of the past couple of years have demonstrated just how quickly day-to-day life can be turned on its head. While mine was thankfully unaffected for the most part, mine became both constrained as well as unpredictable when my partner and I set out a few months ago in search of a new home. We loaded all of our belongings including my tools into a shipping container and hit the road with only what we could fit in the car.

Perhaps we were too naively optimistic about the amount of time it would take to choose a city, plus how competitive the housing market would be just about everywhere. Suffice to say we've been Airbnb hopping for the past few months and while I was hoping to be reunited with my tools and equipment (and other belongings) before this contest concluded... things simply haven't worked out in my favor. As a result I don't have a completed project to submit.

However, I do intend to complete this project. With the exception of the addition of some light cryptography, wireless garage door openers have remained largely the same. It seems that manufacturers have simply decided to leave things as they are. So while there's no need to reinvent the wheel, there's also no need to sit on the sidelines and watch while only Tesla owners get to play with new technologies.

So to Xilinx and Hackster: I apologize for my inability to complete my project at this time. I greatly appreciate you sending me the necessary hardware to compete in the Adaptive Computing Challenge 2021 and will both complete as well as document my project once my situation is more stable!