knobbi

Inspiration

As our team was brainstorming ideas, one of the problems we echoed was pets and kids opening doors as a safety hazard. It is challenging to babyproof or pet-proof doors without making it a hassle for the adults themselves. We realized that this concept also applies to home security with facial recognition, but this is something that already exists--what makes this different? As we were discussing deterring tactics, we mentioned how Tesla car door retraction makes it less attrative for robberies. We sought to incorporate this into our design, knobbi.

What it does

knobbi is a smart door with facial recognition that enables access to only authorized users, with the door knob not extracting when access is denied. This is particularly useful for dementia and alzheimers patients, senior living care, child proofing medical cabinets, and home security.

How we built it

We used OpenCV for the facial recognition, uploading the team members images for training. knobbi has an incorporated camera that provides live feed for real time facial recognition. We laser cut wood for a mockup cabinet to prototype our locking and doorknob retraction mechanisms.

Challenges we ran into

Facial recognition proved to be a challenge as we tested in various brightness and lighting conditions, with false positive cabinet access for wrong person access. This took a lot of iteration to get to a better position. Moreover, integration between the cabinet, the servos, and the electronics for access proved a large challenge as the mechanical interactions behaved unexpectedly.

Accomplishments that we're proud of

Getting a model up and running on a Raspberry Pi, something none of us had prior experience with. Integrating the peripheral camera feed and servos for locking and unlocking within the time constraint that we were challenged with.