you mean dumbed down and automated... it's just brute forcing not picking.. the point of picking is to abuse the tolerances/imperfections of the lock and find the binding pins to speed up the process ;)
LPL would seem to be about 1,000 times faster tho. And the blade-with-wires simply will not go in a lot of zigzagged keyways, like some of Yale's & Best's nastier keyways. Cool proof of concept tho!
Also consider adding rotational compliance between actuator and key. Measuring amount rotated will measure torque. Much more realistic than having encoder inside the door, plus it will reduce fatigue on the keys.
exactly what i was thinking too, you can keep the same setup too just stop rotating the gear if no movement is detected at the encoder after a set angle. make the gear out of nylon or pp and it should last a long time
I was wondering about that kind of thing as well, but think doing it mechanically may be simpler. I was thinking of making the key 15-20cm long and then rotate the far end an 1/8 turn and use the flex in the long key as a spring. At the lock face attach a flag to the key and if it moves the full 1/8 turn you know the core has rotated, if not its just key flex.
Having a key part of the lock picking mechanism inside the house is a big problem... Means that you can already get in, so what need is there to pick the lock? Or have I missed something?
I would say under engineered, as in the realisation of the machine is done by trial and error. When something fails, just go up in size. Where's the engineering? "What will do a lot, will do a little"
The tolerances for the blade is probably fine, but the issue would be the holes for the wire. I doubt there is a service (though I haven't looked) that can do that tolerance for custom jobs, and machining them afterwards would be difficult.
@@akamesama I highly doubt it would be too hard to find a company that can manufacture the blade. The holes are not that small so unless you want to use a difficult material (ex. stainless, some aluminums and cast iron). If it turns out that conventional CNCing is too unreliable something like EDM drilling will have no problems. I'd guess it would cost around $100USD per blade for one-offs.
I have 3d printed metal in sls services. No way the holes will turn as clean. Not at least with the common services like pcbway or jlcpcb services. The minimum feature size is 1mm and it is rough. Had holes about 2mm and had to pass them again with a drill bit.
You don't need to have the whole key being metal, if you have pins on the top and bottom to apply the tension the thin plastic won't have to bear the load.
Exactly, create the first 1mm that goes into the key way and bond the existing 3d printed wire guide to the end of it. This is where the majority of a normal key’s strength for turning the barrel is coming from.
@@Dave_the_Dave or, as it is already overengineered, make the keyblade and its mounting rotational and spring loaded (like a rotational clutch) and enter an electrical endstop springs should be hard enough to rotate the key cylinder but weak enough to soften the sheering moments
The sheer amount of dedication necessary to make this project work is absolutely adorable. The idea seems simple, but there are so many possible failure points in all the details.
As a trained designer/engineer and now current locksmith. I had a similar idea when I first started. This is far more impressive and way wayyyyy cooler than mine. Your editing and video/graphic presentation in your work is remarkable! I will be using your provided files to add to my daily arsenal of bypass tools. Subscribed for sure!
OMG... I drew this up about 1983!!! Of course they say that if you think of anything, a thousand others are thinking the same thing. The difference is, YOU built it! So kudos. Mine was used the same principle, but differed in a couple of details: I drew it with flat wires, which could then "stack" in the key blank, and turn upward to the pins in flat slots. And rather than solenoids, I designed mine with an eccentric cam for each wire... the five or six of which would be in a rotating drum, geared together to quickly go through all possible height combos for the lock... I mean, as you rotated the drum with a crank, the individual cams would also rotate inside, but each one at a slightly different rate, through gearing. Tens of thousands of combinations, I thought, could be gone through in minutes. I also drew it so that different blanks could be attached to the drum, for different locks.
How weird bc I drew up the same idea in 1982!!! So I guess I was first... no big deal 😎 Mine did all the things you were talking about plus it had one of those little knife sharpeners embedded in the back of the machine (manual, of course, not motorized). [SIGH] If only... I could have been even more wealthy than I am today.
@@joew1865 You could have only written that sarcastic, mocking comment if you were incapable of understanding the principles of construction and operation in my description... so you gave yourself away, without even realizing it.
@@proto57yeah, and the comment he wrote is actually copied from something I wrote on a napkin in a bar in 1981 so I also doubt he understands any of the engineering
The thing about this type of attack is that it's the only one that _should_ work if your lock is designed properly. Normal lock picking only works because, in the real world, tight tolerances are really hard to achieve. This type of attack isn't much of a concern; if your lock takes hours to open, it'll hold up to attack far longer than your hinges, your windows, or the security measures on the neighbor's house.
It's very useful if you're trying to open one of your locks. For example, a shed of an inherited house or some small treasure chest. You can always look up the lock online and make a blank.
This is the Lock Picking Engineer, so this video pushed all of my buttons. That was a monumental work of the art of engineering. I've never seen such dedication in a droid before. The slightly easier version is an automated cylinder lock pick, where straight pins could be used instead of curved wires. It looks like you were using a binary counter to set pin heights. Maybe try Gray code to minimize pin movement. You should also be able to reduce the combinations by eliminating large differences between adjacent pin heights which wouldn't be possible when pushing a key into a lock. I loved your Dunning-Kruger graph. 🙂 The entire video was very well done with excellent graphics.
. . . sometimes it is easy to build upon the hard work of a pioneer, once someone takes that initial step, the rest of us can step in and make improvements, thank you sir for creating the initial design
Kind of reminds me of the "key impressioning" technique in which you insert a key blank into a lock, wiggle it, see where the pins left a mark on the blank and file those places 1 step down. Rinse repeat until you got a complete key. But to automate that it would require some sort of measuring capability other comments already mentioned and which is probably quite hard/impossible to achieve.
interesting! if i can throw my unwanted suggestions in: the force behind those wires seems excessive, smaller and weaker linear motors should be sufficient, if all you need is speed and accuracy - another comment mentioned ones from cd drives if you want to turn the key without ruining it, how about an inbuilt tensioning tool? you can turn a springy tensioning tool and measure how far it rotated by the force on the spring, this way you can have some sort of "pick-gun" where you could use it (albeit securely mounted) on an actual in-use lock
Is true linear actuation really needed here? The total travel for the wire between the smallest and largest pin is only a few mm, with a small disc on the driveshaft, attached to the wire, just rotating it a few degrees should move the wire more than linear enough for this use - the larger the disc, the more linear Edit: Building on the original comment, the original stepper motors would probably have been plenty powerful in this configuration - I imagine motors half their size again would still do.
@@Flimzes yeah... the whole project can be built with 5 servos and any cheapo stepper motor. And having the encoder in the door... not adds up for the realistic things as well. Detecting the tension can be done in many different ways, from the piece of plastic which can work as a spring bar, or with pretty cheap TMC220x drivers, they can detect stall. Conclusion: overengineered piece of crap :)
@@felixyasnopolski8571 i was thinking: use the original turner motor, then gear it until it just barely bends the blade, then gear it back a step or two - now it won't break the blade and you just need to detect stalling, no compliant mechanism with sensors. Edit: I think throwing parts at it without considering the problem at hand is underengineering and overbuilding - semantics
Pretty interesting. 1st of all, the metal key is absolutely necessary. Even aluminum ones can get twisted, and you are trying resin/plastic... 2nd of all, you can try to use a base-whatever Gray code, to reduce the number of moved pins.
Applause! Great execution! I've pondered this idea, but with the idea of reading/charting the vibration of the pins being moved. To ideally identify the bumps of the sheer lines. Be it a force sensor on the wire, or just a sensitive microphone against the core. At least to narrow down positions.
Incredible video! Building that robot was impressive on many levels, even if it only worked for bruteforcing. I think to detect, when a pin sets, you would need to make it out of metal to minimize the flexing. Maybe even add an acoustic sensor. Excited for version 3. I hope you get the views you deserve!
Nice work! I feel like not many people understand just how many long, late hours people have to put in to these projects! I'm glad this was on my recommended. Keep up the great work. Also, the animations look very dope!
Strange: I was looking for a steel tube with a key mechanism in the middle that would lock at both ends when I stumbled in here and saw your ROBOT LOCK PICKER. I thought it was pure genius and it's given me loads of other ideas. I'm really grateful. Michael.
This is really cool! If you don't mind my asking, did you ever consider making and testing your prototypes with less pins in the lock? the mechanism is essentially identical for each wire, so once you knew you could fit five wires into a key mould and manipulate them by hand you could've scaled back significantly and saved a lot of money and time by making one or two of the mechanism and testing that.
Also, I have questions about how much you (seemingly) overbuilt this system 🤔 Did you consider using servos rather than stepper motors? For the forces and distances needed and the precision needed here it seems like they'd be fine for this application and much cheaper/lighter/easier to program. If you did test them, what were your findings? Similarly for the key rotation mechanism, did you consider: 1. Having the wires travel through a bowden tube to allow the 90º rotation necessary without moving the whole system as a unit? 2. Decoupling the motor from the keyway with elastic to limit the force put on the key without having to control the motion of the motor itself? It could just turn to a fixed angle and if the key can rotate it will, and if it can't the band will stretch. Again, if you considered/tried these ideas, what were your conclusions/finding? I'm interested in the project but I don't have anywhere near the R&D budget necessary for the number and size of motors and prints you're working with here. So if it's not possible to make something smaller/much cheaper I'd love to know before I try 😅
I did some testing with only a few pins. That is where I got stuck with the encoder problem. It would have caused problems regardless of the number of pins.
I didn't consider servos seriously because of the PID overhead and extra encoders (I also have zero experience with them). The steppers were "good enough". Now if i could have detected pins being set it would have sparked an arms race in making it smaller and faster. The tube is a good idea. Cutting the inertia of twisting the lock is a step in the right direction. I would be mindful though when the tubes twist it will change the height of the pins, I suppose you could compensate in the firmware. I like #2 idea.
I thought for quite a while about making a "Sputnik" from 3D printed parts. So this is very interesting to me. I predict this video will gain a lot of traction. Oh, and thank you for publishing the STL and Fusion files.
This is so cool, I've had this idea in my head for a while but never had the skills to actually make it. Awesome to see somebody else thought the same way and actually went through the effort, and made a dope video out of it too.
Great job! 20 years ago I dreamed of a lot mechanism that used a wire for each tumbler with a 90deg radius and the mechanism with a rotating wheel and springs to randomly go through the patterns. You've gone beyond that simple idea. Only reason why I wanted to use the wires was if one of the tumblers happened to catch in the middle , the rear tumblers would still be cycling whereas with a lock pick tool would be limited if the tumbler actuated at a low height near the middle. Please don't give up, I think you're on to something that will change the entire lock industry. In addition, hopefully will provide you unlimited resources for future development in other areas. I would be surprised if there weren't 3 to 4 novelties worth filing IP for. Hopefully you are protecting the initiative and investment of time sacrificed in your life, I'm sure what you've learned in this project is invaluable, and will be utilized again in the future for something else.
Wow! Amazing video … not sure what impressed me the most, the explainer animations, the amount of skill this required, or your leet keyboarding chops when you CAD!
Aside from your exceptional mechanical AND video editing skills, what’s really cool about these videos, is the way that other people “collectively collaborate and contribute” their ideas to either enhance, or further your original project and ideas!
I have an idea to potentially make the key solve much faster. If you add some sort of sticky thing to the tops of the wires so they can stick to the pins, you would theoretically (budget be willing) be able to pull the pins much faster than the spring would be able to push them down. At that point you just have to pull and twist and the lock should be solved instantly. Might be wrong tho idk seems sound in my head.
Bro about to blow up with this video. Great work on this project. Looking at the pretty large requirements for motors here, I feel like there has to be a way to reduce that number - perhaps you could have the large motor turn a driveshaft, and then a series of simple geared connections that tap power from that drive shaft using a servo or maybe even a solenoid to engage/disengage a clutch? There are also closed loop steppers so you could have the software read back the position of the large rotation, combined with your original load cell design you might be able to sense each sequential pin as well as evaluate false sets - Instead of trying 00001, 00002 - 99998, 99999, you'd test all pin 1 positions, record maximum tumbler deflection for a given force (or record a curve or position:force), reset p1 to 0 and then test all pin 2 positions - slowly build a "best guess". If that guess is say, 55555 - back to brute force but all combos within 1 or 2 of that baseline (testing 1 step would be 44444 to 66666), reduces your tests from 10⁵ to somewhere between 3⁵ (1 step) and 5⁵ (2 steps). Depending on how accurate of a closed stepper and load cell you get... I bet this accuracy could be brought lower and lower. If you also added another load cell on the wires to "feel" resistance there... Well it isnt going to help complexity but more sensor data will probably help reduce your Nⁿ search even further! 😂 Again, tremendous work, great video - _love_ the editing and visuals youve put together. Toss in an explosion and you're rivalling early Mythbusters production quality. Looking forward to seeing you grow!
Fun project, some very nice engineering. One little remark: When using a setscrew to fix a gear (or pulley, for that matter) onto a shaft, the setscrew should be on the opposite side of the flat (the round part of the shaft), in effect pushing the flat parts together. The setscrew isn't there to prevent rotation, it's there to ensure close fitment of the flat faces to transmit torque across the parts.
Cool idea. Sounds fun. A couple of things to consider (plus a bonus idea) 1) Have you considered using the linear steppers they use in hard drives. They are super cheap 2) Since your key wire guide design seems pretty stable, have you considered buying a metal one online (like Xometry or Protolabs)? 3) Another interesting idea for making a metal key wire guide would be to cast one using a low temperature casting metal and a high temperature wire that will not bond to the metal (like titanium) to form the wire paths.
this thing looks like some NASA thruster or something with all the actuators converging to one point. Wild! Awesome project! I can imagine how fiddly and painful getting this to work must have been. Take your well deserved break!
Some remarks I had during the video. Might help with version 3. 1) You need two motors. A big one for pushing the wires and rotating the lock and a small one for "switching gears". You rarely move more than one motor at a time anyway. 2) 13:44 - Have you tried using two keys to make one? Like molding two parts and joining them because one mold would be too hard to pull off. Or you could order a custom metal part. 3) This might actually become a useful product! You could duplicate almost any key if you know its template. It does no damage, is reusable and even works on cylindrical locks.
You are simply my hero!! The frustration you had at every obstacle was palpable and I know how that is with any project. But how you persevered and moved forward... is sadly something I am working on! What a cool project!
How do you not have a hundred times as many subscribers, at least? Your production is solid, and your project is interesting! There is the problem of this not being able to pick a lock in situ, but I imagine that this is what version 3 is for.
This is awesome! I had thought about something very similar to this, but hadnt had a way to try to make it. Just a thought though, had you considered using some sort of metal insert into your key blade to make it more robust? We lockpickers use windshield wiper blade inserts a lot for various tools including tensioners, maybe having one of the really thin ones at the top in the plastic or the bottom edge would help?
Yeah. I roughly prototype'd that metal blade I mentioned in the video. A slip of the cutoff wheel ruined it. It was a moot point though I realized while debating on making a second one. Since the plastic blade was "good enough". The whole design still had that inherit slow flaw. In other words, a metal / stronger blade wouldn't have improved the design only saved me time from rewiring it due to broken blades.
@@SparksandCode I meant a hollow spot in the plastic you could insert a metal piece to make the plastic take less of a twisting force. We do that when 3d printing keys out of pla, or when trying to make a few other tools to keep the plastic from deforming as much. I'll take out the calipers to give you an idea of the dimensions I'm talking about when I get up in the morning. 👍🏻
@@SparksandCode You only need to have small metal bits at the opening of the keyway, top and bottom, which are connected to a metal disk, which is the thing you apply force to. The plastic stuff gets no torsion, it's just along for the ride. PS: I'm the BobCat in "Picking The Bridge" in the New Yorker. Don't do what we did. ;)
Just finished the vid and when I subscribed I noticed u have less than 2k subs?!??! I was expecting 100s of thousands. Let's get this man to 10k guys 💪
Question: since the lock is now integrated in the middle, the robot is limited to locks removed from their door. If you could get all the components on one side, it could actually open locked doors. You could still use an optical encoder, but you need a secondary signal to compare against. You need a way to compare the key blade rotation to the system rotation. So essentially 2 sensors.
. . . sometimes it is easy to build upon the hard work of a pioneer, once someone takes the initial step, it is easy for the rest of us to step in and make improvements, thank you sir for creating the initial design
I’m massively into engineering and science and UA-cams algorithm suggested this video, so you’re doing something right 😂. Keep up the good work, you’ll get there, pick that damn lock! Subscribed! 👍🏻
Great project vlog! I do wonder how much pain you would have spared yourself if you had decided to try some different resins early on. The fast resins tend to be very brittle, but a lot of functional resins behave more like ABS. There might also be some room for a hybrid design, where you print the fine details but attach those to a metal key base somehow; the load-bearing part isn't the same as the fine-detail part...
FYI, your "key" breaking issue looks to be because of stress risers from a square interface with the main circular hub/part (see 3:49) since they all appear to break at almost exactly the same spot. If you add a small fillet, even a tiny one, its strength will increase significantly.
More modern stepper drivers such as the TMC2209, are capable of measuring resistance on the stepper motor, which would be useful for sensing if the lock has been picked or not, the force is set in firmware/software(depends on how you want to implement it), so you can easily tune the force to be just right where it will be able to turn the key once unlocked but won't be able to shear the plastic key.
You could also use a smaller stepper motor with a worm reduction(metal or printed), this could provide a lot of torque without taking up a lot of space, and you also get to rotate the motor 90°.
Absolutely respect the hustle, what a huge project for a relatively small UA-cam channel, I couldn't presently dare the volume of effort you have gone into and quality of video! great work bud, enjoying the watch! (at the time of this comment you had 3.27K subs, keen to see the growth!)
I'm going to contest your encoder simply because you put it on the other side of the lock. If anything, you need some form of force feedback to determine whether that cylinder is free or not. Simple load cell on a spiral spring would do the job. In fact a load cell can determine when each pin has cleared as well
Nice work! I considered making a picking tool years ago based on the same principles as yours but I never got around to doing it. I was trying to do something the size of a cordless screwdriver using one motor to sequentially nudge the wires - almost like an automated rake that moved the pins directly just like yours. I'm impressed that you did it. I gave up because I couldn't make it portable. If you record the pin heights, you'll have a lock key decoder, sorts like like a lichee pick, but automated.
Picking the right resin will impact the brittleness a lot, especially water washable. Also make sure the resin isn’t too cold, the bottle will give you the recommended temps.
I wonder if there are any easily measurable forms of feedback you could make use of to speed up the process? For example, is there a distinct change in electrical resistance, or perhaps a particular sound or resonance when the pins are aligned properly, that could be detected by a multimeter or sensitive microphone. Maybe there's a usable signal (electrical or audible) the pick could generate, which would be modulated in a detectable way. It seems feasible, to me.
For a future version, instead of using an optical encoder you can measure the current needed to turn the motor that's rotating the key. Sorta how modern 3d printers do nozzle bed leveling, you would do a few light turns before starting to get a base line of a locked state and then do a momentary test every time you get a wire into a new location.
This is cool to see. 21 weeks ago, apparently, I commented to a friend on FB "Now I want to build a key with tiny flutes running through it, and spring wires hooked to geared stepper motors, and then a robot can just quickly run through the 150k combos for me," because I have a cabinet with a very good lock I cannot for the life of me pick. I actually seriously considered doing this exact project˘which is a lie; I did crazy projects when I was 23, but now I'm twice that, and don't have the same drive to dive into endless efforts like this). I also actually considered printing every possible key, but I don't know which key yet, and yeah, those combinatorics... I figured I could fit 50 keys on the plate per print, 4 hours, 15 minutes per print, using 82g of PLA, and I'd need 117,649 keys, so only 2,353 full-plate prints, taking almost exactly 10k hours (10000.165!). I'd also need 192,864g of filament, which is only 193 reels, so about $2600 with ANYCUBIC 4-packs ($13/reel). I figured I'd only need to do about 1500 runs before I landed on the key, so about 6,375 hours, or 797 8-hour printing shifts. Not too bad :)
Great video. Taking a break is a great idea. Your focus will be much better when you return. Thanks for sharing your journey. Don't give up. I'm looking forward to the next chapter when you return.
I would replace the large helical gears with 'standard ones.' Helical gears introduce axial forces which, considering that most of your components are 3D printed, can lead to unnecessary bending and twisting of your construction. Yes, I understand that helical gears provide more contact surface, resulting in a smoother transmission, but I believe it's worth considering alternatives.
Should’ve filed two keys down to half the original thickness then used a CNC mill to make passages for the wires on one of both halves (pending wire gauge) before welding the two halves together
Very nice project! Why not just spin the lock instead of the key assembly? I was also wondering if you could add a distance sensor for each wire so that no test turn is necessary.
@@SparksandCodeI receive similar comments on my projects, too. Once I get going on an idea, it's full speed ahead, and I forget to ask myself if there are alternative solutions.
Make a safe-cracking robot. A group at MIT made an example, once. I almost had to make one to get into a safe (until its mechanism jammed, entirely). One of the funnest parts of a combo-lock solver is figuring out just how (and how much) you can speed up the search. You can spin the dial really fast, but you can't just stop immediately or the discs inside will keep going, so you have to program in some acceleration into your stepper code.
i wonder if you could decouple the key from the motors by using plastic tubing as ducts for the wires. you don't need to rotate the key much, and the flex of plastic tubes should be more than enough. lighter key will require a lighter mount, lighter mount could be driven with a lighter motor, lighter motor won't chew through plastic as quickly.
Nice take 😄. Have you thougth about using "Gray code" to improve the search speed (at least a little). This might be considered "premature optimisation" but if you are familiar with Gray codes the required implementation time, effort and number of lines of code to change should be rather insignificant.
Another side benefit to using this tool is you can begin to map out the locking styles of different lock manufacturers. So, for example, Master locks May only push their pins in a certain distance and or only have a certain degree of variance. All very hard information to get without having a tool that's digital. Kind of scary when you think about it...
A fun item for a DnD game. The rogue in a party has a skeleton key that can open any common lock. However, it's huge, it's not silent and it's cursed with being slow. The picking ability is a part of the curse. If the curse is removed, the thing becomes useless.
I don’t usually leave comments, but as a fan of creators like Stuff Made Here, Marius Hornberger, Jeff Gerling, Ivan Miranda, and many others in the tech and engineering space, I’m thrilled to have discovered your channel. Truly underrated-I hope more people discover this gem. You've absolutely earned a new subscriber. Your videos beautifully illustrate the complexity and time it takes to turn an idea from a concept into a tangible creation. The dedication and meticulous attention to detail you show in facing these challenges are commendable. I appreciate your perseverance in using trial and error, a powerful learning tool that many abandon after too many failures. I eagerly look forward to more videos of this caliber. One aspect I’d love to see more of is your coding approach. You don't need to go into every detail, but more on the structure and reasoning behind your decisions would be great. A high-level explanation of your thought process, combined with your insights into engineering, would be truly enlightening-pure gold, platinum, even unobtanium! Comments alone really can’t convey the depth of my appreciation and admiration for your work. Please keep the videos coming!
It wouldn't. It's a neat concept, but the only lockpicking it affords the user is brute force guessing the pin combination. There's zero market for a tool like that when any lock-related problem has a much quicker and usually cheaper solution.
The wires don´t need tension, but that doesn´t mean you cannot use it. To put less pressure on the plastic of the key itself, it could have a slightly wider top part made out of metal, kind of functioning like a tensioning tool. Then, the mechanical stress of the rotation is mostly applied to this metal part.
If you use Nitinol memory wires you can control them like little motors using simple electrical signals. We are working on a skin tight exoskeleton that has thousands of thin .5mm Nitinol memory wires running through the entire suit with pickup along all major muscle groups to fire the wires when needed. In short if you go to lift something very heavy, simply try to lift it and once your muscle groups in your arm slightly tighten the pickups read the electrical signal through your skin and tell a micro controller to fire the corresponding Nitinol memory wires thus greatly assisting the weight your lifting. I should mention Nitinol is made from titanium and nickel so its super strong!
You could do the same thing with two motors. One connected to a linear actuator and the other connected to a wheel with holes for each wire (think revolver but wire instead of bullets). This way you can upgrade this setup to as many wires as needed without increasing the motor count beyond 2
Why stepper motors instead of servos, it's not as if the stroke length needed to be long and wire wouldn't be sufficient over any length. Servos would give adequate control and force reducing build size and removing the need for linear conversion or sensing position. A metal cast key head would prevent breakage, I assume through excessive twisting as the lock is tested.
I had to rewind a couple times to see if it actually worked. Maybe you could make it more apparent, for example: have the monitor in view showing different combinations and a success result. Would be great to have as the format of your story telling builds up to the climax, and then... "wait, what happened?" Felt like the payoff wasn't there for us as viewers. Anyway, glad I found your channel, subscribed! Looking forward to more!
Oh man, you've got some great design skills! We live in a fantastic time where 3D printing and electronics availability allow smart creators to do stuff like this. Keep it up, I'm subscribing!
For version 3 you could maybe do the encoder on the key side and have some kind of torque sensing to determine when pins are set. Motor current sensing could work pretty well for that. That would let you sense that little give when a pin sets properly, and it might let you generalize the robot to work on actual doors.
An alloy you can cast against plastic is cerrotru. Make a negative for the blade as well and reprint the key with only the tubes then put the two together and pour the alloy to fill. Maybe T slot to keep it together. You are probably already familiar with Rose and Wood's metal, LPL just had some for a kit, well... cerrotru is stronger and dimensionally stable (alloys grow and shrink when cooled and over time, depending), no lead either. Not kidding about low temp, I use it to inlay wood like they used to with pewter. I only use blue painter's tape for a dam and it does not burn it or the wood. It's also just called 281⁰f, that's it solidus And melting point (there can be ranges by alloy) so it casts at 350 to not chill halfway down. It could probably work.
Question, if the pins above are the same length, the ones that don't move with the tumbler when unlocked, why not push all pins to full with the wires on low torque so they only go to max and no further, then have the machine draw the wires back the distance of those top pins? Since the wires will adjust and slip to the right lengths due to the low torque, when they are pulled back the right amount they are all aligned since usually that second pin above the varying pins are all uniform in length. I know there's other things like springs and what not, but under full load most of that is generally going to be the same length with the only variations being the interchangeable pins used to interface with the key. But if you just have the robot push all pins to full and have it so the motor torque can only push them to full and no further, drawing them back the right distance will align the key pins. Granted the easy way to fix this is to have the second pins above be different lengths as well, but what I've noticed from most lock picking videos is those pins tend to be the same length. So instead of developing an attack just for the pins used in a traditional pick, why not develop an attack to use the second pins as part of the attack?
You can buy the tiniest little linear servos. They are inexpensive. It is my understanding that the moving parts there come from a camera lens focusing mechanism. The 28byj shouldn't be this slow either, i don't know why you're struggling to turn them faster. But i wouldn't use them due to inherent backlash. You can also buy precision linear stepper mechanisms intended for cheap optical disk drives.
You could use a some sort of spring like one of those spiral springs in watches to apply torque until it hits a microswitch. That allows you to mechanically limit the torque instead of relying on the precision of the gearing and encoder.
Great project. Im 5 minutes in and the obvious question! Why didn't you use RC servo motors? They are servo motors, plenty of control, small low current fast, and plenty strong enough to push a wire with a short lever arm after all the wire doesn't need to move very far.
That's awesome! I invite you to look into using pulse generators to speed up the stepper motors, using smaller gear cogs and more gears for additional accuracy and torque, and using impregnated resin. But I'm nobody, just an armchair coach who has never played the game. Love to watch! 👍
5:32 You could save weight if you'd use a single motor and do something like a punchcard system. With pressured air to move the wires this could make the device even smaller. Similar to a barrel organ. Another method to determin a set pin could be to put a microphone to the lock and write an algorithm for the sharp sound, a set pin makes. Since you use wires instead of picks, the noise from picking the lock would be minimal.
"This is the lockpicking robot, and what I have for you today is..."
"Today I'm outside this commenters house, and were gonna see if we can get in their backdoor..."
@@JohnHilton-dz4mi "oh fuck yeah spread it" ( ͡° ͜ʖ ͡°)
@@MagicGumable oh shit. Lol
lpl
@@MagicGumable🤢
Lockpicking lawyer getting automated
He needs to see this
LPL: they took my jerb!
you mean dumbed down and automated... it's just brute forcing not picking.. the point of picking is to abuse the tolerances/imperfections of the lock and find the binding pins to speed up the process ;)
@@robertjung8929 Indeed!!!
LPL would seem to be about 1,000 times faster tho. And the blade-with-wires simply will not go in a lot of zigzagged keyways, like some of Yale's & Best's nastier keyways. Cool proof of concept tho!
Also consider adding rotational compliance between actuator and key. Measuring amount rotated will measure torque. Much more realistic than having encoder inside the door, plus it will reduce fatigue on the keys.
I was thinking the same thing.
exactly what i was thinking too, you can keep the same setup too just stop rotating the gear if no movement is detected at the encoder after a set angle. make the gear out of nylon or pp and it should last a long time
Compliant mechanisms are just super cool, and 3d printers are almost made for them lol
I was wondering about that kind of thing as well, but think doing it mechanically may be simpler. I was thinking of making the key 15-20cm long and then rotate the far end an 1/8 turn and use the flex in the long key as a spring. At the lock face attach a flag to the key and if it moves the full 1/8 turn you know the core has rotated, if not its just key flex.
Having a key part of the lock picking mechanism inside the house is a big problem... Means that you can already get in, so what need is there to pick the lock?
Or have I missed something?
Painfully overengineered. I like it.
Lol, the damn size of the motors alone.... to drive an iron wire forwards......
I would say under engineered, as in the realisation of the machine is done by trial and error. When something fails, just go up in size. Where's the engineering?
"What will do a lot, will do a little"
It needs more lube
Once you have a part finalized, like that resin keyblade, send it to a company, like shapeways, to have it metal SLSed or whatever, for full strength
The tolerances for the blade is probably fine, but the issue would be the holes for the wire. I doubt there is a service (though I haven't looked) that can do that tolerance for custom jobs, and machining them afterwards would be difficult.
@@akamesama I highly doubt it would be too hard to find a company that can manufacture the blade. The holes are not that small so unless you want to use a difficult material (ex. stainless, some aluminums and cast iron). If it turns out that conventional CNCing is too unreliable something like EDM drilling will have no problems. I'd guess it would cost around $100USD per blade for one-offs.
Yes! And then instead of a complex encoder system, you can use basic limit switches.
I have 3d printed metal in sls services. No way the holes will turn as clean. Not at least with the common services like pcbway or jlcpcb services. The minimum feature size is 1mm and it is rough. Had holes about 2mm and had to pass them again with a drill bit.
Use a piece of spring steel alongside the "key" to reinforce and use as a tension wrench. Use plastic for wire feed only
"Yo dude I forgot my key, what do I do?"
"Hold on..."
*Whips out Lock Pickenator 9000*
Whi-pie!
wtf is whipes bro 😭😭
The robot wiped out... how? Are we talking about Cool Whip™? I'm so confused by your joke!
*whips
Waited his whole life for this moment
You don't need to have the whole key being metal, if you have pins on the top and bottom to apply the tension the thin plastic won't have to bear the load.
Exactly, create the first 1mm that goes into the key way and bond the existing 3d printed wire guide to the end of it.
This is where the majority of a normal key’s strength for turning the barrel is coming from.
But really, it isn't hard to do this entire key on metal.
lost pla casting would make it easy to just cast the whole thing in 1 go
A small tension bar would suffice
@@Dave_the_Dave or, as it is already overengineered, make the keyblade and its mounting rotational and spring loaded (like a rotational clutch) and enter an electrical endstop
springs should be hard enough to rotate the key cylinder but weak enough to soften the sheering moments
The algorithm has decided this video is gonna make it.
The mans ability to persevere in spite of the project being completely useless and inefficient is amazing.
The sheer amount of dedication necessary to make this project work is absolutely adorable. The idea seems simple, but there are so many possible failure points in all the details.
i mean ... this IS a pick. thats a beautiful project.
It's the Pick of Destiny
As a trained designer/engineer and now current locksmith. I had a similar idea when I first started.
This is far more impressive and way wayyyyy cooler than mine.
Your editing and video/graphic presentation in your work is remarkable!
I will be using your provided files to add to my daily arsenal of bypass tools.
Subscribed for sure!
As a 30’year Licensed Locksmith, this is nonsense.
He says he “picked the lock while watching the pins in a lock with a cutaway! SMH!
Nonsense!
Lockpicking lawyer laughing from a dark corner menacingly.😂
OMG... I drew this up about 1983!!! Of course they say that if you think of anything, a thousand others are thinking the same thing. The difference is, YOU built it! So kudos.
Mine was used the same principle, but differed in a couple of details: I drew it with flat wires, which could then "stack" in the key blank, and turn upward to the pins in flat slots. And rather than solenoids, I designed mine with an eccentric cam for each wire... the five or six of which would be in a rotating drum, geared together to quickly go through all possible height combos for the lock... I mean, as you rotated the drum with a crank, the individual cams would also rotate inside, but each one at a slightly different rate, through gearing. Tens of thousands of combinations, I thought, could be gone through in minutes.
I also drew it so that different blanks could be attached to the drum, for different locks.
How weird bc I drew up the same idea in 1982!!! So I guess I was first... no big deal 😎 Mine did all the things you were talking about plus it had one of those little knife sharpeners embedded in the back of the machine (manual, of course, not motorized). [SIGH] If only... I could have been even more wealthy than I am today.
If you care about progress more than copyright, send a copy of your designs to him. Make sure to add descriptions. He might make it a reality.
@@joew1865 You could have only written that sarcastic, mocking comment if you were incapable of understanding the principles of construction and operation in my description... so you gave yourself away, without even realizing it.
@@proto57yeah, and the comment he wrote is actually copied from something I wrote on a napkin in a bar in 1981 so I also doubt he understands any of the engineering
underrated youtuber
Fr
Not for long
Enjoy the ride, folks! We got on the train relatively early!
The thing about this type of attack is that it's the only one that _should_ work if your lock is designed properly. Normal lock picking only works because, in the real world, tight tolerances are really hard to achieve.
This type of attack isn't much of a concern; if your lock takes hours to open, it'll hold up to attack far longer than your hinges, your windows, or the security measures on the neighbor's house.
It's very useful if you're trying to open one of your locks. For example, a shed of an inherited house or some small treasure chest. You can always look up the lock online and make a blank.
@@Uthael_Kileanea Right, but in the real world, a Lishi tool is cheaper and much faster. And makes it easier to cut a new key once you do pick it.
@@Uthael_Kileaneait’s pretty useful if you need to pick a lock that you can access the back side of to mount an encoder
This is the type of account, that is keeping UA-cam from becoming a social media cesspool.
phenomenal work!!
This is the Lock Picking Engineer, so this video pushed all of my buttons. That was a monumental work of the art of engineering. I've never seen such dedication in a droid before. The slightly easier version is an automated cylinder lock pick, where straight pins could be used instead of curved wires. It looks like you were using a binary counter to set pin heights. Maybe try Gray code to minimize pin movement. You should also be able to reduce the combinations by eliminating large differences between adjacent pin heights which wouldn't be possible when pushing a key into a lock.
I loved your Dunning-Kruger graph. 🙂 The entire video was very well done with excellent graphics.
. . . sometimes it is easy to build upon the hard work of a pioneer, once someone takes that initial step, the rest of us can step in and make improvements, thank you sir for creating the initial design
Kind of reminds me of the "key impressioning" technique in which you insert a key blank into a lock, wiggle it, see where the pins left a mark on the blank and file those places 1 step down. Rinse repeat until you got a complete key. But to automate that it would require some sort of measuring capability other comments already mentioned and which is probably quite hard/impossible to achieve.
interesting!
if i can throw my unwanted suggestions in: the force behind those wires seems excessive, smaller and weaker linear motors should be sufficient, if all you need is speed and accuracy - another comment mentioned ones from cd drives
if you want to turn the key without ruining it, how about an inbuilt tensioning tool? you can turn a springy tensioning tool and measure how far it rotated by the force on the spring, this way you can have some sort of "pick-gun" where you could use it (albeit securely mounted) on an actual in-use lock
Is true linear actuation really needed here?
The total travel for the wire between the smallest and largest pin is only a few mm, with a small disc on the driveshaft, attached to the wire, just rotating it a few degrees should move the wire more than linear enough for this use - the larger the disc, the more linear
Edit: Building on the original comment, the original stepper motors would probably have been plenty powerful in this configuration - I imagine motors half their size again would still do.
@@Flimzes yeah... the whole project can be built with 5 servos and any cheapo stepper motor. And having the encoder in the door... not adds up for the realistic things as well. Detecting the tension can be done in many different ways, from the piece of plastic which can work as a spring bar, or with pretty cheap TMC220x drivers, they can detect stall. Conclusion: overengineered piece of crap :)
@@felixyasnopolski8571 i was thinking: use the original turner motor, then gear it until it just barely bends the blade, then gear it back a step or two - now it won't break the blade and you just need to detect stalling, no compliant mechanism with sensors.
Edit: I think throwing parts at it without considering the problem at hand is underengineering and overbuilding - semantics
@@Flimzes true! i feel like spending less time implementing and more time designing/ideating could have benefitted the project
What about using Bowden tubes to feed the wires through for reduce friction and use servos and linkages to drive them?
Pretty interesting.
1st of all, the metal key is absolutely necessary. Even aluminum ones can get twisted, and you are trying resin/plastic...
2nd of all, you can try to use a base-whatever Gray code, to reduce the number of moved pins.
Applause! Great execution! I've pondered this idea, but with the idea of reading/charting the vibration of the pins being moved. To ideally identify the bumps of the sheer lines. Be it a force sensor on the wire, or just a sensitive microphone against the core. At least to narrow down positions.
I like that microphone idea
Incredible video! Building that robot was impressive on many levels, even if it only worked for bruteforcing. I think to detect, when a pin sets, you would need to make it out of metal to minimize the flexing. Maybe even add an acoustic sensor. Excited for version 3.
I hope you get the views you deserve!
Glad you enjoyed it!
Nice work! I feel like not many people understand just how many long, late hours people have to put in to these projects! I'm glad this was on my recommended. Keep up the great work. Also, the animations look very dope!
Also I just noticed, we have the EXACT same "Box of Shame"! Same usps box and size, filled with failed prints haha
Strange: I was looking for a steel tube with a key mechanism in the middle that would lock at both ends when I stumbled in here and saw your ROBOT LOCK PICKER. I thought it was pure genius and it's given me loads of other ideas. I'm really grateful. Michael.
How do you not have a million subscribers?! This is awesome!
Because he wrangled a massive amount or effort into a brute force machine. It's not impressive.
@@kasparroosalu my guy, the editing, presentation, and effort put into this video is impressive.
@@kasparroosaluas opposed to what? An artistic lock picking machine? Lmfao
@@amogusenjoyer lock pick gun? a set of picks? a good wave-raking?
Amazing work. I can't imagine the amount of suffer you have had to endure for it. Well done for pushing through.
This is really cool!
If you don't mind my asking, did you ever consider making and testing your prototypes with less pins in the lock? the mechanism is essentially identical for each wire, so once you knew you could fit five wires into a key mould and manipulate them by hand you could've scaled back significantly and saved a lot of money and time by making one or two of the mechanism and testing that.
Also, I have questions about how much you (seemingly) overbuilt this system 🤔
Did you consider using servos rather than stepper motors? For the forces and distances needed and the precision needed here it seems like they'd be fine for this application and much cheaper/lighter/easier to program. If you did test them, what were your findings?
Similarly for the key rotation mechanism, did you consider:
1. Having the wires travel through a bowden tube to allow the 90º rotation necessary without moving the whole system as a unit?
2. Decoupling the motor from the keyway with elastic to limit the force put on the key without having to control the motion of the motor itself? It could just turn to a fixed angle and if the key can rotate it will, and if it can't the band will stretch.
Again, if you considered/tried these ideas, what were your conclusions/finding? I'm interested in the project but I don't have anywhere near the R&D budget necessary for the number and size of motors and prints you're working with here. So if it's not possible to make something smaller/much cheaper I'd love to know before I try 😅
I did some testing with only a few pins. That is where I got stuck with the encoder problem. It would have caused problems regardless of the number of pins.
I didn't consider servos seriously because of the PID overhead and extra encoders (I also have zero experience with them). The steppers were "good enough".
Now if i could have detected pins being set it would have sparked an arms race in making it smaller and faster.
The tube is a good idea. Cutting the inertia of twisting the lock is a step in the right direction. I would be mindful though when the tubes twist it will change the height of the pins, I suppose you could compensate in the firmware.
I like #2 idea.
I just realized you meant servo not dc servo (brushless DC motor with encoder). Yeah I always forget about those. That's a good idea
I thought for quite a while about making a "Sputnik" from 3D printed parts. So this is very interesting to me. I predict this video will gain a lot of traction. Oh, and thank you for publishing the STL and Fusion files.
The repo in the description has the STL for the manual one.
Let me know how it works.
This is so cool, I've had this idea in my head for a while but never had the skills to actually make it. Awesome to see somebody else thought the same way and actually went through the effort, and made a dope video out of it too.
Great job! 20 years ago I dreamed of a lot mechanism that used a wire for each tumbler with a 90deg radius and the mechanism with a rotating wheel and springs to randomly go through the patterns. You've gone beyond that simple idea. Only reason why I wanted to use the wires was if one of the tumblers happened to catch in the middle , the rear tumblers would still be cycling whereas with a lock pick tool would be limited if the tumbler actuated at a low height near the middle. Please don't give up, I think you're on to something that will change the entire lock industry. In addition, hopefully will provide you unlimited resources for future development in other areas. I would be surprised if there weren't 3 to 4 novelties worth filing IP for. Hopefully you are protecting the initiative and investment of time sacrificed in your life, I'm sure what you've learned in this project is invaluable, and will be utilized again in the future for something else.
Wow! Amazing video … not sure what impressed me the most, the explainer animations, the amount of skill this required, or your leet keyboarding chops when you CAD!
Dude's killin' it and only 1k subs? Shows you how messed up the algo is... Congrats on your triumph fellow maker!
Appreciate it!
New on my feed, liked and subscribed. Looking forward to the next vid!
Aside from your exceptional mechanical AND video editing skills, what’s really cool about these videos, is the way that other people “collectively collaborate and contribute” their ideas to either enhance, or further your original project and ideas!
I have an idea to potentially make the key solve much faster. If you add some sort of sticky thing to the tops of the wires so they can stick to the pins, you would theoretically (budget be willing) be able to pull the pins much faster than the spring would be able to push them down. At that point you just have to pull and twist and the lock should be solved instantly. Might be wrong tho idk seems sound in my head.
Bro about to blow up with this video.
Great work on this project. Looking at the pretty large requirements for motors here, I feel like there has to be a way to reduce that number - perhaps you could have the large motor turn a driveshaft, and then a series of simple geared connections that tap power from that drive shaft using a servo or maybe even a solenoid to engage/disengage a clutch? There are also closed loop steppers so you could have the software read back the position of the large rotation, combined with your original load cell design you might be able to sense each sequential pin as well as evaluate false sets -
Instead of trying 00001, 00002 - 99998, 99999, you'd test all pin 1 positions, record maximum tumbler deflection for a given force (or record a curve or position:force), reset p1 to 0 and then test all pin 2 positions - slowly build a "best guess". If that guess is say, 55555 - back to brute force but all combos within 1 or 2 of that baseline (testing 1 step would be 44444 to 66666), reduces your tests from 10⁵ to somewhere between 3⁵ (1 step) and 5⁵ (2 steps). Depending on how accurate of a closed stepper and load cell you get... I bet this accuracy could be brought lower and lower.
If you also added another load cell on the wires to "feel" resistance there... Well it isnt going to help complexity but more sensor data will probably help reduce your Nⁿ search even further! 😂
Again, tremendous work, great video - _love_ the editing and visuals youve put together. Toss in an explosion and you're rivalling early Mythbusters production quality. Looking forward to seeing you grow!
Fun project, some very nice engineering. One little remark: When using a setscrew to fix a gear (or pulley, for that matter) onto a shaft, the setscrew should be on the opposite side of the flat (the round part of the shaft), in effect pushing the flat parts together. The setscrew isn't there to prevent rotation, it's there to ensure close fitment of the flat faces to transmit torque across the parts.
Cool idea. Sounds fun.
A couple of things to consider (plus a bonus idea)
1) Have you considered using the linear steppers they use in hard drives. They are super cheap
2) Since your key wire guide design seems pretty stable, have you considered buying a metal one online (like Xometry or Protolabs)?
3) Another interesting idea for making a metal key wire guide would be to cast one using a low temperature casting metal and a high temperature wire that will not bond to the metal (like titanium) to form the wire paths.
this thing looks like some NASA thruster or something with all the actuators converging to one point. Wild! Awesome project! I can imagine how fiddly and painful getting this to work must have been. Take your well deserved break!
Some remarks I had during the video. Might help with version 3.
1) You need two motors. A big one for pushing the wires and rotating the lock and a small one for "switching gears". You rarely move more than one motor at a time anyway.
2) 13:44 - Have you tried using two keys to make one? Like molding two parts and joining them because one mold would be too hard to pull off. Or you could order a custom metal part.
3) This might actually become a useful product! You could duplicate almost any key if you know its template. It does no damage, is reusable and even works on cylindrical locks.
Wow... the amount of effort is incredible... good job!
Thanks a lot!
You are simply my hero!! The frustration you had at every obstacle was palpable and I know how that is with any project. But how you persevered and moved forward... is sadly something I am working on! What a cool project!
As a fellow lock picker, i think this is bloody awesome!!
You definitely have my subscription
How do you not have a hundred times as many subscribers, at least? Your production is solid, and your project is interesting! There is the problem of this not being able to pick a lock in situ, but I imagine that this is what version 3 is for.
I like how you filmed and then edited all the footage together with funny bits and cool renderings. Great job.
Glad you enjoyed it!
This is awesome! I had thought about something very similar to this, but hadnt had a way to try to make it. Just a thought though, had you considered using some sort of metal insert into your key blade to make it more robust? We lockpickers use windshield wiper blade inserts a lot for various tools including tensioners, maybe having one of the really thin ones at the top in the plastic or the bottom edge would help?
Yeah. I roughly prototype'd that metal blade I mentioned in the video. A slip of the cutoff wheel ruined it.
It was a moot point though I realized while debating on making a second one. Since the plastic blade was "good enough". The whole design still had that inherit slow flaw. In other words, a metal / stronger blade wouldn't have improved the design only saved me time from rewiring it due to broken blades.
@@SparksandCode I meant a hollow spot in the plastic you could insert a metal piece to make the plastic take less of a twisting force. We do that when 3d printing keys out of pla, or when trying to make a few other tools to keep the plastic from deforming as much. I'll take out the calipers to give you an idea of the dimensions I'm talking about when I get up in the morning. 👍🏻
@@SparksandCode You only need to have small metal bits at the opening of the keyway, top and bottom, which are connected to a metal disk, which is the thing you apply force to. The plastic stuff gets no torsion, it's just along for the ride.
PS: I'm the BobCat in "Picking The Bridge" in the New Yorker. Don't do what we did. ;)
Just finished the vid and when I subscribed I noticed u have less than 2k subs?!??! I was expecting 100s of thousands. Let's get this man to 10k guys 💪
Maybe one day!
Same! Although atm its like 7k subs. Waaaay too low!
Question: since the lock is now integrated in the middle, the robot is limited to locks removed from their door. If you could get all the components on one side, it could actually open locked doors. You could still use an optical encoder, but you need a secondary signal to compare against. You need a way to compare the key blade rotation to the system rotation. So essentially 2 sensors.
I think you can get your key reprinted in metal from one of those services. then you can switch back to the load cell. awesome work
. . . sometimes it is easy to build upon the hard work of a pioneer, once someone takes the initial step, it is easy for the rest of us to step in and make improvements, thank you sir for creating the initial design
Amazing video.
Superb graphics and production, love your story telling and inclusion of faults and fixes, Thankyou. You’ve gained a sub :-)
Pretty awesome build. I'd love to see a more optimised version with much smaller/faster motors.
That manual pick is pretty incredible too.
I’m massively into engineering and science and UA-cams algorithm suggested this video, so you’re doing something right 😂. Keep up the good work, you’ll get there, pick that damn lock!
Subscribed! 👍🏻
Great project vlog! I do wonder how much pain you would have spared yourself if you had decided to try some different resins early on. The fast resins tend to be very brittle, but a lot of functional resins behave more like ABS. There might also be some room for a hybrid design, where you print the fine details but attach those to a metal key base somehow; the load-bearing part isn't the same as the fine-detail part...
FYI, your "key" breaking issue looks to be because of stress risers from a square interface with the main circular hub/part (see 3:49) since they all appear to break at almost exactly the same spot. If you add a small fillet, even a tiny one, its strength will increase significantly.
More modern stepper drivers such as the TMC2209, are capable of measuring resistance on the stepper motor, which would be useful for sensing if the lock has been picked or not, the force is set in firmware/software(depends on how you want to implement it), so you can easily tune the force to be just right where it will be able to turn the key once unlocked but won't be able to shear the plastic key.
Resistance is not the most correct word, but I think you get what I meant.
You could also use a smaller stepper motor with a worm reduction(metal or printed), this could provide a lot of torque without taking up a lot of space, and you also get to rotate the motor 90°.
Absolutely respect the hustle, what a huge project for a relatively small UA-cam channel, I couldn't presently dare the volume of effort you have gone into and quality of video! great work bud, enjoying the watch! (at the time of this comment you had 3.27K subs, keen to see the growth!)
Glad you enjoyed it!
I'm going to contest your encoder simply because you put it on the other side of the lock.
If anything, you need some form of force feedback to determine whether that cylinder is free or not.
Simple load cell on a spiral spring would do the job.
In fact a load cell can determine when each pin has cleared as well
Nice work! I considered making a picking tool years ago based on the same principles as yours but I never got around to doing it. I was trying to do something the size of a cordless screwdriver using one motor to sequentially nudge the wires - almost like an automated rake that moved the pins directly just like yours.
I'm impressed that you did it. I gave up because I couldn't make it portable.
If you record the pin heights, you'll have a lock key decoder, sorts like like a lichee pick, but automated.
Picking the right resin will impact the brittleness a lot, especially water washable. Also make sure the resin isn’t too cold, the bottle will give you the recommended temps.
I wonder if there are any easily measurable forms of feedback you could make use of to speed up the process? For example, is there a distinct change in electrical resistance, or perhaps a particular sound or resonance when the pins are aligned properly, that could be detected by a multimeter or sensitive microphone. Maybe there's a usable signal (electrical or audible) the pick could generate, which would be modulated in a detectable way. It seems feasible, to me.
For a future version, instead of using an optical encoder you can measure the current needed to turn the motor that's rotating the key. Sorta how modern 3d printers do nozzle bed leveling, you would do a few light turns before starting to get a base line of a locked state and then do a momentary test every time you get a wire into a new location.
This is cool to see. 21 weeks ago, apparently, I commented to a friend on FB "Now I want to build a key with tiny flutes running through it, and spring wires hooked to geared stepper motors, and then a robot can just quickly run through the 150k combos for me," because I have a cabinet with a very good lock I cannot for the life of me pick. I actually seriously considered doing this exact project˘which is a lie; I did crazy projects when I was 23, but now I'm twice that, and don't have the same drive to dive into endless efforts like this). I also actually considered printing every possible key, but I don't know which key yet, and yeah, those combinatorics... I figured I could fit 50 keys on the plate per print, 4 hours, 15 minutes per print, using 82g of PLA, and I'd need 117,649 keys, so only 2,353 full-plate prints, taking almost exactly 10k hours (10000.165!). I'd also need 192,864g of filament, which is only 193 reels, so about $2600 with ANYCUBIC 4-packs ($13/reel). I figured I'd only need to do about 1500 runs before I landed on the key, so about 6,375 hours, or 797 8-hour printing shifts. Not too bad :)
Great video. Taking a break is a great idea. Your focus will be much better when you return. Thanks for sharing your journey. Don't give up. I'm looking forward to the next chapter when you return.
You can print the key part with Siraya Tech Blu or another tough resin, or have one of those 3d printing services print you one in metal.
I've got Siraya Tech Blu in my Amazon cart right now ;)
@@SparksandCode Looks like Siraya Tech Build might be better? You might want to check out some reviews before you order.
will do
I would replace the large helical gears with 'standard ones.' Helical gears introduce axial forces which, considering that most of your components are 3D printed, can lead to unnecessary bending and twisting of your construction. Yes, I understand that helical gears provide more contact surface, resulting in a smoother transmission, but I believe it's worth considering alternatives.
Should’ve filed two keys down to half the original thickness then used a CNC mill to make passages for the wires on one of both halves (pending wire gauge) before welding the two halves together
I am soo happy you didn’t give up especially when you wanted to
Thank you
Very nice project! Why not just spin the lock instead of the key assembly? I was also wondering if you could add a distance sensor for each wire so that no test turn is necessary.
You know… spinning the lock never occurred to me.
@@SparksandCodeI receive similar comments on my projects, too. Once I get going on an idea, it's full speed ahead, and I forget to ask myself if there are alternative solutions.
Exactly.
I’ve been nerd sniped by a few of these comments that have creeped into my rough idea for v3
@@SparksandCode Nice. I'll be on the lookout for the next version.
Make a safe-cracking robot. A group at MIT made an example, once. I almost had to make one to get into a safe (until its mechanism jammed, entirely). One of the funnest parts of a combo-lock solver is figuring out just how (and how much) you can speed up the search. You can spin the dial really fast, but you can't just stop immediately or the discs inside will keep going, so you have to program in some acceleration into your stepper code.
can't believe he put over a year of his life into one 16 minute video.
i wonder if you could decouple the key from the motors by using plastic tubing as ducts for the wires. you don't need to rotate the key much, and the flex of plastic tubes should be more than enough. lighter key will require a lighter mount, lighter mount could be driven with a lighter motor, lighter motor won't chew through plastic as quickly.
Nice take 😄.
Have you thougth about using "Gray code" to improve the search speed (at least a little). This might be considered "premature optimisation" but if you are familiar with Gray codes the required implementation time, effort and number of lines of code to change should be rather insignificant.
Another side benefit to using this tool is you can begin to map out the locking styles of different lock manufacturers. So, for example, Master locks May only push their pins in a certain distance and or only have a certain degree of variance. All very hard information to get without having a tool that's digital. Kind of scary when you think about it...
A fun item for a DnD game. The rogue in a party has a skeleton key that can open any common lock. However, it's huge, it's not silent and it's cursed with being slow. The picking ability is a part of the curse. If the curse is removed, the thing becomes useless.
I don’t usually leave comments, but as a fan of creators like Stuff Made Here, Marius Hornberger, Jeff Gerling, Ivan Miranda, and many others in the tech and engineering space, I’m thrilled to have discovered your channel.
Truly underrated-I hope more people discover this gem. You've absolutely earned a new subscriber.
Your videos beautifully illustrate the complexity and time it takes to turn an idea from a concept into a tangible creation. The dedication and meticulous attention to detail you show in facing these challenges are commendable.
I appreciate your perseverance in using trial and error, a powerful learning tool that many abandon after too many failures. I eagerly look forward to more videos of this caliber.
One aspect I’d love to see more of is your coding approach. You don't need to go into every detail, but more on the structure and reasoning behind your decisions would be great. A high-level explanation of your thought process, combined with your insights into engineering, would be truly enlightening-pure gold, platinum, even unobtanium!
Comments alone really can’t convey the depth of my appreciation and admiration for your work. Please keep the videos coming!
So this is stuffmadehere with the voice of Jeremy fielding.
I love the choice of music in this, very suspenseful.
The manual version you made first has some serious potential. With metal construction that would sell like hot cakes
It wouldn't. It's a neat concept, but the only lockpicking it affords the user is brute force guessing the pin combination. There's zero market for a tool like that when any lock-related problem has a much quicker and usually cheaper solution.
The wires don´t need tension, but that doesn´t mean you cannot use it. To put less pressure on the plastic of the key itself, it could have a slightly wider top part made out of metal, kind of functioning like a tensioning tool. Then, the mechanical stress of the rotation is mostly applied to this metal part.
If you use Nitinol memory wires you can control them like little motors using simple electrical signals. We are working on a skin tight exoskeleton that has thousands of thin .5mm Nitinol memory wires running through the entire suit with pickup along all major muscle groups to fire the wires when needed. In short if you go to lift something very heavy, simply try to lift it and once your muscle groups in your arm slightly tighten the pickups read the electrical signal through your skin and tell a micro controller to fire the corresponding Nitinol memory wires thus greatly assisting the weight your lifting. I should mention Nitinol is made from titanium and nickel so its super strong!
MT6701 or AS5600 are great magnetic encoders and can be mounted directly to the back of the stepper with the magnet on the shaft. Great project!
You could do the same thing with two motors. One connected to a linear actuator and the other connected to a wheel with holes for each wire (think revolver but wire instead of bullets). This way you can upgrade this setup to as many wires as needed without increasing the motor count beyond 2
Why stepper motors instead of servos, it's not as if the stroke length needed to be long and wire wouldn't be sufficient over any length. Servos would give adequate control and force reducing build size and removing the need for linear conversion or sensing position.
A metal cast key head would prevent breakage, I assume through excessive twisting as the lock is tested.
I had to rewind a couple times to see if it actually worked. Maybe you could make it more apparent, for example: have the monitor in view showing different combinations and a success result.
Would be great to have as the format of your story telling builds up to the climax, and then... "wait, what happened?" Felt like the payoff wasn't there for us as viewers.
Anyway, glad I found your channel, subscribed! Looking forward to more!
Oh man, you've got some great design skills! We live in a fantastic time where 3D printing and electronics availability allow smart creators to do stuff like this. Keep it up, I'm subscribing!
For version 3 you could maybe do the encoder on the key side and have some kind of torque sensing to determine when pins are set. Motor current sensing could work pretty well for that.
That would let you sense that little give when a pin sets properly, and it might let you generalize the robot to work on actual doors.
An alloy you can cast against plastic is cerrotru. Make a negative for the blade as well and reprint the key with only the tubes then put the two together and pour the alloy to fill. Maybe T slot to keep it together.
You are probably already familiar with Rose and Wood's metal, LPL just had some for a kit, well... cerrotru is stronger and dimensionally stable (alloys grow and shrink when cooled and over time, depending), no lead either.
Not kidding about low temp, I use it to inlay wood like they used to with pewter. I only use blue painter's tape for a dam and it does not burn it or the wood. It's also just called 281⁰f, that's it solidus And melting point (there can be ranges by alloy) so it casts at 350 to not chill halfway down.
It could probably work.
Question, if the pins above are the same length, the ones that don't move with the tumbler when unlocked, why not push all pins to full with the wires on low torque so they only go to max and no further, then have the machine draw the wires back the distance of those top pins? Since the wires will adjust and slip to the right lengths due to the low torque, when they are pulled back the right amount they are all aligned since usually that second pin above the varying pins are all uniform in length. I know there's other things like springs and what not, but under full load most of that is generally going to be the same length with the only variations being the interchangeable pins used to interface with the key. But if you just have the robot push all pins to full and have it so the motor torque can only push them to full and no further, drawing them back the right distance will align the key pins.
Granted the easy way to fix this is to have the second pins above be different lengths as well, but what I've noticed from most lock picking videos is those pins tend to be the same length. So instead of developing an attack just for the pins used in a traditional pick, why not develop an attack to use the second pins as part of the attack?
You can buy the tiniest little linear servos. They are inexpensive. It is my understanding that the moving parts there come from a camera lens focusing mechanism.
The 28byj shouldn't be this slow either, i don't know why you're struggling to turn them faster. But i wouldn't use them due to inherent backlash. You can also buy precision linear stepper mechanisms intended for cheap optical disk drives.
Love seeing projects like this. Keep it up!
I think you should place the rotary encoder on the main shaft and use a clutch system that prevents the blade from being turned further than possible.
these graphics are absolutely fire; great work on the project
I am just watching this to get rid of it from being suggested down my throat all the time
You could use a some sort of spring like one of those spiral springs in watches to apply torque until it hits a microswitch. That allows you to mechanically limit the torque instead of relying on the precision of the gearing and encoder.
Great project. Im 5 minutes in and the obvious question! Why didn't you use RC servo motors? They are servo motors, plenty of control, small low current fast, and plenty strong enough to push a wire with a short lever arm after all the wire doesn't need to move very far.
That's awesome! I invite you to look into using pulse generators to speed up the stepper motors, using smaller gear cogs and more gears for additional accuracy and torque, and using impregnated resin. But I'm nobody, just an armchair coach who has never played the game. Love to watch! 👍
thanks, I conceptualised this machine about 40 years ago, glad to see someone made it at last - nice work.
Glad I could help!
I conceptualized the flying car when I was six. So like, whenever somebody does all of funding, research, and development for that... You're welcome.
5:32 You could save weight if you'd use a single motor and do something like a punchcard system. With pressured air to move the wires this could make the device even smaller.
Similar to a barrel organ.
Another method to determin a set pin could be to put a microphone to the lock and write an algorithm for the sharp sound, a set pin makes. Since you use wires instead of picks, the noise from picking the lock would be minimal.
And then one night, I find the pre science-future invention that vindicates the sci-fi concept of an auto-pick. Brilliant man.