This Robot Picks Locks, If You’re Very Patient

We all know the Hollywood trope of picking a lock with a paperclip, and while it certainly is doable, most reputable locks require slightly more sophisticated tools to pick effectively. That’s not to say that wire is off the table for locksports, though, as this cool lock-picking robot demonstrates.

The basics behind [Sparks and Code]’s design are pretty simple. Locks are picked by pushing pins up inside the cylinder until they line up with the shear plane, allowing the cylinder to turn. Normally this is done a pin at a time with a specialized tool and with a slight bit of torque on the cylinder. Here, tough, thin, stiff wires passing through tiny holes in a blade shaped to fit the keyway are used to push all the pins up at once, eliminating the need to keep tension on the cylinder to hold pins in place.

Sounds simple, but in practice, this looks like it was a nightmare. Getting five wires to fit into the keyway and guiding them to each pin wasn’t easy, nor was powering the linear actuators that slide the wires in and out. Applying torque to the lock was a chore too; even though tension isn’t needed to retain picked pins, the cylinder still needs to rotate, which means moving the whole picking assembly. But the biggest problem by far seems to be the fragility of the blade that goes into the keyway. SLA might not be the best choice here; perhaps the blade could be made from two thin pieces of aluminum with channels milled on their faces and then assembled face-to-face.

The robot works, albeit very slowly. This isn’t [Sparks and Code]’s first foray into robot lock picking. His previous version attempted to mimic how a human would pick a lock, so this is really thinking outside the box.

17 thoughts on “This Robot Picks Locks, If You’re Very Patient

    1. I think you could have saved a great deal of time by “paying upfront” and making a metal key blank from the get go. Just because you have a hammer (3d printer) doesn’t mean every problem should be treated as a nail.

  1. Stepper motors are soo alluring to people coming from Arduino, but once you have encoders they make little $ sense. Like that safe cracking one using super expensive stepper
    I dont quite get going from load cell to optical encoder with a fragile plastic key. Why not measure power draw of a motor to estimate load?
    Hand picking is all about feeling the resistance, I would imagine robot should do the same.

  2. [Sparks and code] did very well here in recreating the iconic lockpick tool called a Sputnik. The original tool predates the patent (US5172578A) by about a decade, where it was first used by an inventive burglar in Europe. You can still buy these tools from some locksmithing tool suppliers, but they gladly ask several thousand euros for just one of these lockpicks, which in turn only works on one keyway of one brand.

    From the video, it’s unclear if the creator of this YT video was aware of the Sputnik. Mostly as you can use it to pick a lock instead of the brute force attempt shown in the video. This tool works especially well if the lock contains one standard pin and several spool pins (As was the case for Abus back then).

    You first set the standard pin to the correct height, and then wobble the lockpicking left and right while pushing up one spool pin at the time. The wobble decreases while you push the pin and the wobble increases again when the pin is set. Oliver Diderichsen’s video shows it doesn’t take more than a minute if you know what you are doing.

    The creator shared the files, for which he gets full points. The manual version, shown off in blue, was excluded from the files, sadly. That’s the one the lockpicking community likes the most. We would like to use a metal key though, it’s not too difficult to create with a bit of patience :)

  3. I remember about 30 years ago there was this guy who made a safe cracker machine that could hear the tumblers fall, and open any safe in less than 30 seconds. Security officials were freaking out. I think they got\ the guy to destroy his little safe cracker machine. But several other versions by others has appeared since then.

  4. This reminds of the turbodecoder, which completely mechanically decodes and opens a lock in seconds. They are very effective, but easy to damage the blade if used by inexperienced locksmiths.

  5. Neat, and since you know exactly how far each pin has to go in to open the lock with that data you should easily be able to recreate a key for the lock you’ve just picked.

  6. Why not put an encoder/firce sensor on the wires for the pins.
    Lets call the pin down as the 0 point.
    Push the wire all the way in until it can go any more then release it backl to 0
    do this on all poins and you will get an estimation as to how long each one is.
    Once you know how long the pins are you can pretty much push them up until they clear the barrel and allow it to turn.

  7. The thing with the wires.. That brought back old memories. I worked at a place that did a lot of contract work for DEC. One of them was repairing the logic boards from the LA36 printers. I repaired thousands of them. In one of the other buildings they did mechanical assembly and they manufactured the print head assemblies. I never paid much attention to them outside of the solenoids that drove the print head wires were a ways back from the head and it used wires, though they did not have a bend in them. If you could work out a mechanical system like in the valves in a car, in the key so all the wires only ran in straight lines, I suspect you could cycle them orders of magnitude faster.

    One interesting stupid trick you could pull on those printers, is the has an encoder to show the print head position, you could take the printing print head and jerk it ahead and the printer would usually keep up with it. They were capable of going a lot faster than they went.

    After I left that job, I saw those things in odd places for the next 20 years or more. It has been a while since I have seen one in the wild now, but I suspect they still have some uses. One thing they were great at was making copies. I think the last time I saw one was in a muffler shop and the thing was cranking out 4 copy forms.

    1. Other options include some device to turn the force 90°. Rack and pinion maybe, or make it function more like a lockpick with 4-5 ‘blades’ each working one pin. Maybe skip the ‘key blade’ all together to save complexity. Either grab into the keyway from top and or bottom like a tension tool, or turn the ‘picking blades’ as a whole.

      I think this system might benefit from an organized cracking table using most common pinnings first to least common.

  8. I may be missing something, but this looks like it still has trouble finding out when a pin is set, and falls back to brute forcing a too-large search space. The almost-as-simple device that you can just cheaply buy for whatever kind of lock you like is a lishi tool. That has one pick, but the pin locations are labeled as well as the depths, so you can take the numbers and cut a key just like with this one. Like it’s always been since the invention of this kind of lock, it relies on how one pin at a time will usually bear most of the tension, and then it will release and fall on another once the first pin is set whether falsely or correctly. Sure, security pins will get you, but the professionals (LPL) seem to just return and continue picking when that happens.

  9. I made a manual version of this almost 20 years ago.
    I guarantee it would have been easier for them if they hadn’t wasted so much time 3d printing this garbage/toy.

    It was simple to fit the piano wire into that space.
    The hard part is moving the wire in precise increments.

    My solution was using really small zip-ties that happened to have the teeth spaced the same distance as each pin length.

    It worked.
    It is also WAY worse than a modern LeeChi tool.
    In fact, automating a LeeChi tool would have been a WAY more useful way to do it.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.