This robot picks locks using brains instead of brawn

A physical lock, like what secures your front door, has a finite and calculable number of combinations, just like a digital keypad does. There are a set number of pins in the lock and each can be one of a set number of lengths. Each of those numbers varies based on manufacturer and model, but that information is easy to find. As a result, it is possible to brute force such a lock by trying all of the combinations. But that’s time-consuming, so this robot built by Sparks and Code works smarter instead of harder.

In this case, “smarter” meant deducing the pin lengths instead of trying all of the combinations in sequence. The technique used by the robot to perform that deduction is what makes this project so interesting.

Each pin in the lock has a spring that pushes it down against the key’s edge. Typically, the springs for all of the pins are identical. But because the pins are of different lengths, the force required to push a pin to a specific height (above an imaginary reference line) varies. By measuring that force and comparing it against reference measurements for known pin lengths, the robot can guess the length of the pin and therefore the correct “combination” (the key bittings).

The robot has an Arduino Nano board that measures the spring force by pushing a wire into the lock with a servo motor. That servo motor mounts onto a load cell, which outputs a signal proportional to the force on the servo motor and therefore the wire and therefore the spring. The robot has such an assembly for each of the five lock pins.

This idea, while very clever, proved to be difficult to implement in the real world. Sparks and Code struggled to get accurate measurements and had to rely on collecting several measurements to average. Even that didn’t work well on many of the pins. 

But the concept is still intriguing and we hope to see Sparks and Code continue with the development.