A facial-recognition system to prevent getting locked out

This project started with a problem: the design of the dorm room doors allowed students to exit and close the door while it remained locked, causing many, including myself, to lock themselves out. After a few iterations, I was able to design a reliable, secure system to unlock my door using a facial recognition library in Python running on a Raspberry Pi, and a stepper motor controlled by an Arduino that moves a custom 3D-printed doorknob. I also implemented a soft-close system since the door became too heavy to reliably close on its own. I had to mount everything with command strips and duct tape, since I didn’t own the door, but this system could be integrated much more cleanly as a more permanent solution, which I am currently working on.

  • The proof-of-concept build made completely using things I had lying around. I had to extend the doorknob to be able to move it using a "winch" setup.

  • A slightly refined system using a raspberry pi

  • I added a screen, a new camera which fit inside the peephole, cooling fans on the raspberry pi, and a doorbell outside.

  • With the move to a stepper motor and pulley, the design looks much more refined. Mounting has also been reworked.

  • The final design before leaving the dorm. All critical parts have been reprinted using my new 3D printer, and a soft-close system has been added.

Final hardware test

Proof-of-concept hardware test.

Soft close test