Keerat Singh (keerats2@illinois.edu, ECE 110) and Prudhvie Gudapati

Introduction

Statement of Purpose

As busy college students, we are often rushing out to various activities without remembering to make sure everything is the way it should be. An example of this is forgetting to lock the door when we are rushing to our 8 AM classes 5 minutes before class starts. Our proposal to counteract this is to build a facial/speech recognition lock that the does not require turning. Along with the physical lock, we plan to create a mobile phone application that can be used to control the lock away from home.

Background Research

We are motivated to work on this project because it makes our lives more convenient in the long run. Also, this project will make our living spaces more secure as accidentally leaving a door unlocked can cause valuable items to be stolen very quickly. When we brainstormed different projects, our motivations were the same. We wanted to make our lives more convenient and render our memory as unnecessary as possible. Other projects we thought of were combo lock breakers, skateboards that will come to you with a simple command, and bike directions using augmented reality. Our proposed project is similar to our other projects ideas in the use of basic software coupled with hardware such as Arduinos.

There are many products available that use facial recognition to lock and unlock door, such as Corum¹ Security’s CS-100. That is obviously far more of a finished product than what we will make, and the CS-100 is equipped with RFID, something that checks for the location of your phone, and many other measures on top of just facial recognition. Because our project will be designed for use in our dorms, which would already require the person getting in to pass through the security of the building, there is less of a risk even if it does fail. There are also products like the Nest doorbell and Amazon’s DropCam that are part of a smart home network, but again, our project would not be a commercial product that needs to interface with other products, except for our phones.

On the other side of the spectrum, many people have posted instructions online for DIY projects that are similar to what we are trying to accomplish, like this post on O’Reilly² and this post on Adafruit³. However, the Adafurit project uses Microsoft Azure, and we have decided to use Amazon’s AWS Rekognition service instead. The O’Reilly project uses Rekognition, but it interfaces with an AMazon Echo and is designed to let people in the house already know who is at the door rather than react to the person at the door by opening it.

Design Details

Block Diagram / Flow Chart

System Overview

Our project has two inputs that will scan one’s face and send it to the Raspberry Pi. The Pi will then send the information to the Amazon Rekognition service where it will run its facial recognition algorithms to decide whether the face is part of the software’s database. Rekognition will then send the final decision to the Pi which will decide whether or not to unlock the door. A feature on the phone app will be a button that will automatically lock/unlock the door from a far distance. This information will be sent to the Pi via WiFi (probably).

Parts

-Motor
-Raspberry pi
-Lock (similar mechanism)
-Normal hardware toolkit
-Camera
-SD Card/USB (if necessary)

Possible Challenges

Figuring out how to use facial/speech recognition software
Learning to how develop mobile applications
Software and Hardware connection (application to board/pi)
Ability to finish in one semester

ECE 110/120 Honors Lab Section : Facial Recognition Lock