Your experience with making things, building projects, and hacking hardware is directly proportional to the amount of components you have on hand; as our experience grows, so do our space and storage requirements, it seems. [Danh Trinh] must have decades of experience, because his StorageBot robotic parts drawer is as awesomely absurd as it is clever and useful.
At first glance, StorageBot just looks like a bunch of small parts drawers mounted to the wall with LED strips along the top and side. The magic happens when [Danh] walks up to the wall-mounted laptop and commands StorageBot to find a component with his voice. A video is worth a thousand words, so you might as well head to the video for the best description available
To get the StorageBot to listen to his voice, [Danh] downloaded Microsoft’s speech recognition SDK and built a VB app to turn his voice into a location of what drawer contains the part he requested. Once StorageBot finds the row and column of the requested part, a pair of stepper motors behind the wall of parts drawers swing into action. Soon enough, the drawer containing the requested part pops out, and [Danh] can go about his business building more awesome stuff.
Because a few paragraphs can’t convey exactly how cool [Danh]’s StorageBot is, take a look at the videos after the break.
[youtube=http://www.youtube.com/watch?v=LbRsGL2lCqA&w=470]
[youtube=http://www.youtube.com/watch?v=uIaPYj399VI&w=470]
Eleven!
I wish him good luck in his bidding for the Shop Bot prize. He’s demonstrated ability and ambition and clear use for the Shop Bot should he win. I’ve had a similar scheme in mind for a long time but at a much smaller scale. I’m sure many of us Makers have. In the end, Makers can’t make more time to match their whimsy. So it’s just cool to see someone get it done.
Sick
I saw one of this guy’s videos a while back. I think he is proof that with true genius always comes a little weird.
“Listen now – screws – assortment – small”. Stand exactly 1 m away, with no background noise. Press a key before you start :)
This is a nice project, but it looks like it requires significant *human* training and adjustment to use properly. It is also probably speaker dependent.
Thumbs up overall. Great effort.
With all likelyhood, he’s a hardware guy and not a software guy.
I don’t think the 1m distance is significant, but that definitely depends on the microphone quality. The MS Voice Recognition is pretty good, but depending on how complex your queries are there isn’t a single voice recognition package out there that is perfect every time.
It’s a really awesome project. Maybe next he’ll set something up to automatically organize those “spare parts” bins he has ;)
TBH, this goes right back to his reuse point. He could just as well be using some drop-in voice recognition software. I thought that The Point is to reuse existing designs to enable modular complexity increases.
Just make sure you get them all back in the right spot when you put them back.
why would you be standing much further than about 1m away from the unit anyway? I presume as it’s got a laptop controlling the shebang that he can ssh in and request his parts from somewhere else if he wants? Could be pretty simple to use a mobile phone as a mobile terminal to use anywhere.
i was thinking of doing something similar for resistor color codes :)
I think the value of this project would be tremendously increased if he could come up with modular bins, so that as he requires more space, he merely has to clip another bin to the end of the row. This is obviously limited by the space, and the structure would need to be incorporated into the modular bins. I think that a modular version of what he has would make a viable commercial product.
Great looking project though!
Cool project. The voice interface seems very clunky. It’s a good proof of concept, but I think it detracted from an otherwise awesome project.
An improvement that would work with the voice interactions he showed would be to wire an led to each bin. As the search is refined, it could light up all of the matching boxes. Using LEDs to mark the matching bins (and remove the auto open) would make the build cheaper and open the door for it to be a more modular system.
That was my thought!
And additionally, when it’s not locating parts, the entire array could be playing Conway’s Game of Life ;)
Also would think LED’s would be a step up here. Opening the drawers is cool but slower than LED’s.
Additionally, and I don’t know if someone has mentioned this, maybe having BoM feature where the system will search for your parts on a BoM, and highlight them via the LED’s.
This is certainly something I’m putting on my build list.
I agree, modular bins would be a great asset, as would the addition of QR coded bins and a camera so that putting them back in the wrong spot wouldn’t matter- perhaps it could be modified to resort the bins as well for quicker access to the most used parts and slower access to the least used(or least accessed parts) additionally the rate of speed could be better, all in all a great project itself though and nicely done.
pharmacies and hardware stores will love this.
Actually pharmacies is a good idea, with a little modification it could even fill the bottle with the number of pills requested.
Such things exist. Here’s an example. You could conceivably build a custom dispensing library for something like this, fill your SMD devices into empty gelatin capsules, and have it bottle up a parts kit for you.
http://www.kirbylester.com/pharmacy-products-KL60.htm
http://www.scriptpro.com/
OK, however, simply using the colour code led grid thing would suffice. He is standing anyway, so a little colour on the led strips to indicate location, but now he has to burn juice on the motors.
I would also address the relocating issue. A simple barcode/rfid waved past the unit and it again indicates where to put the box back. I would probably order a bunch of boxes in one go, grab then all and sit them on the workbench. As an other mentioned, training the user is critical, so make it simple.
Wow. I was just reading about this concept in Cory Doctorow’s book Makers, and was wondering if it would ever be a reality. Dahn has made my day.
As far as I can tell the only thing missing is a way to deliver the parts to the bench. Hopefully the StorageBot never goes berserk, shoving all the drawer out the front
I just cleaned up my work area. i would hate to see what you just happened :P
As amazing as this is, I think the best part is that I was looking at these same bins on Harbor Freight (On sale for $13.49!) because I really need parts bins, and LOTS OF THEM, but was hesitant because most of the reviews said they were kind of flimsy!
That machine is absolutely mind-blowing though. Wow. And what he had to say about the ‘maker’ movement is really spot-on, that’s exactly the way I feel!
The robotic and voice aspects add a lot of cool/drool factor. I understand this is entered in a competition, and I hope he wins! But speaking practically, I think I’d tire of the inaccuracy/inflexibility of speech recognition, and waiting for the mechanism to open a drawer when I can generally find and open it faster.
The searchable database is the best part of this. I have my own, which is now used for lots of stuff in my home. It was easy enough to make in Access using two related tables – Containers and Items. Each container is labelled physically and in the database with a unique name, and can contain items or other nested containers. Nesting allows me to, for example, know that an item is in a particular small partitioned organizer, in a larger Rubbermaid box, even on a particular shelf in a particular room (if I set up those as containers too). This allows me to find anything quickly in any conceivable organizational hierarchy, using any containers I want. For drawers as depicted in this project, I add a row/column location to the item. I can also attach optional pictures to each item, which frequently comes in handy.
One more idea. Though parts drawers are great, they’re not terribly space efficient. To keep a particular screw separate, you have to dedicate an entire drawer partition to it, even if you have only two of those screws; the end result is that a “full” organizer is usually mostly empty. Labelled, nesting Ziploc bags of various sizes can be more space efficient, certainly cheaper, and still fairly fast to locate parts in. For example, nested four deep: Hardware/Machine Screws/6-32/1.5″.
Chris would you like to share the database framework? I have the same system but just a spreadsheet and pics folder.
As previously stated, this is nice for a proof of concept, but the truth is that there is a lot of room for improvement. Speech-recognition is always drool-worthy, but isn’t very practical. In the amount of time it takes to go through all the commands, one could have just gone through all the likely drawers and found the part manually, especially if the bins were labeled on the outside. I’ve never met a piece of fast voice-recognition software.
I do really like the automation of this process though. As a previous commentor said, for a viable product, he could remove the stepper motor portion, and just use an LED on the drawer itself. Or, if automated drawer opening is required, use something quicker, like soldenoids or a pulley/conveyor belt system (although much more complicated).
Some improvements:
-Make the search happen via touchscreen on that computer. Type in a standard description of what you’re looking for, then have the computer match it to your inventory. This way, you could create a list of items that you need, maybe a BoM for a project, and it could locate all of the items at once. If you were using this system to build a project, it would take forever to get all the parts (“Please wait while I reset”).
-You could link each finished project’s BoM with a QR or bar code in the event that you need to duplicate it in the future. Scan the code and the computer automatically locates all the parts in the BoM. This could also be implemented with a menu of some kind onthe touchscreen, but would be less cool.
-Combine this with the Project Management portion of Mouser, and you could implement the previous comment even easier. Also find a way to make it work with this:
http://hackaday.com/2012/03/12/ecdb-offers-a-simple-intuitive-way-to-keep-your-electronic-inventory-organized/
-Use RGB LEDs so that you have a visible cue when you are running low on a specific part. Yellow for low, red for one left, or completely empty.
Tea, Earl Grey, Hot!
Up Vote!
I understand that this is rather a demo made to inspire people and it’s great at that. To be practical it needs to be faster. I know from experience that sometimes 5 seconds of unnecessary wait time are more than enough to provoke some rage. I think by a visual cue I could locate any box much faster than this robot. Given that these are my boxes and not a shared warehouse, I could probably locate stuff even without any visual cue just by memory.
Instead of slow steppers in an open-loop CNC-like arrangement, it should use a pair of fast servos: the precision will be limited to the size of a cell and positional feedback should be read from the rails. Thus this machine can be made scary fast.
i think everyone dissing on the voice-rec being lanky/overkill is severely discounting the lovely convenience of not having to break one’s train-of-thought/stride in the midst of a project only to trivially hunt and peck for the right part. The moment you start worrying about these “technical” details is when your inner tony stark begins to die.
a case of inventor vs. engineer?
Actually a picture is worth a thousand words and if the the rate of 24 FPS is standard then we can possibly assume that this video is in 24 FPS and therefore 24 FPS X 625 seconds (10:25) = 15,000. Therefore this video is worth 15,000 words. Did i do all that right? I’m bad at math so i wouldn’t be surprised to hear that i did it wrong.
:)
15000 words would only be for 15 frames or less than one second. The proper formula would be 24 FPS x 1000 words/frame x 625 seconds = 15 million words.