Let's Make Robots!

Scarab-A Roomba for dog poop in a fenced-in yard

Navigate via touch sensors and sweep dog poop up into a disposable plastic shopping bag.

This project was named the Scarab not just for its resemblence to the dung beetle but for its functional similarity to that animal.  In principal this machine is intended to wander around a fenced in yard and act as an autonomous turd-combine that deposits its treasure in a standard disposable grocery bag.  In practice, picking up turd is a little more complicated than I've been able to master. The solution may be as simple as using a lower gear-down on the auger, or as complex as putting a 12v shop vac on it. The platform is a converted RC Tank base with two bumper switches and a simple offset-delayed electromechanical H-bridge pair for brains.  Eventually I'd like to do a microprocessor version that can home in on a charging base (but the usability of IR outdoors worries me) or eventually use GPS to limit the unit to a yard that isn't fenced in.

Update 8/28/2012

I was going through my old files and discovered some media to add that, when I first posted this, I didn't have the LMR savvy/experience to include.  Hope yall like it.  As I mentioned, I originally called it the "Turd Hurler," so if that shows up in any files, well, you get the picture...

Just about as simple as you can get and still act like a Roomba.  Here's what it looks like on the premade PCB.  Not quite up to the level of acuity of BlueBeta or Sanc04 on similar perfboards, but it works.  

This is where I began my practice of color-coding the wiring as much as possible, so where you see red it's probably Vcc and black is the ground rail.  All this fits into the water-proof pelican case on the ass of the unit (the yellow box) and the wires have little filed channels that you have to push them into to get the damned thing to close.  The terminal strip is on top of the battery compartment and it connects the "Brain" to the motors, battery, etc.

That's a pretty big file-I think if you click on it you'll see the unscaled version.  I laid it out full-scale in AI CS3 to get a feel for the scale and how the pieces would fit together before the tank came to me.  Obviously a lot of things changed.

Of course, it's a WIP probably forever.  If I ever get back to it and fix the problem with its failure to pick up poop I might also upgrade it to μcontroller brains and ultrasonics (probably overkill and useless against chain link fences without the bumpers.)

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

This is something I've been thinking about (I love my dogs but would like the poo collection to be automated) My idea on poo detection would be to use IR cameras to detect the "hot" spots from several angles and thus work out it's position, I do like your take on collection with a rotating rake. I'm currently working with arduino's in the garden (I've got 12V running round the garden and am starting to use ESP2886 Wifi chips for communications, an ir detector on a pan and tilt unit sweeping till it finds a hot spot and then triangulating the results should detect a fresh poo and give a close approcimation to it's location !

nice build how messy does that get.... awesome idea tho

I need one of these! This is great, love the design. Looking forward to seeing more from this.


If you have a dog, then cleaning the yard is a crap job (pun intended).  If you get this working I'd love to see / steal it!


Instead of having the scoop so high, wouldn't it be better to have it almost at ground level (that is if you regularly mow your lawn).


Yes it is. Of course since the dogs lived with my fiancé before I met her and I moved into her house, she feels obligated to do this and all of the yard work. That's probably to motivating me to get it working...

You're right about the other thing too-the scoop, which is actually more of a set of rakes on an axel, does brush the ground four times with every revolution of the auger

Could you not use a methane gas sensor to detect poop?

Thanks Bajdi, that is some good thinking!
I think most digestive methane is released before waste elimination and not so much from the pooh itself. Also, either of the theoretical sensors would be useful only while the poop was still fairly fresh, which would make the ag-combine approach to collection, well... kinda gross.. Not that the idea behind the Scarab is exactly flower picking. But a laser methane detector style unit like they use when the Utility company comes to investigate a leak was what I had in mind. It turns out that the absorption frequency (!) of the main gas that makes pooh stinky, Hydrogen Sulfide, is too far into the ultra violet to be useful in an open setting.
Also, I don't really know if it needs to sense the pooh to pick it up. Look at a Roomba. It doesn't vacuum the way you and I do, looking for visible floor crud and going to it. It just vacuums everywhere and eventually gets it all. It takes longer than you or I would but it also doesn't have any other job or hobbies/favorite tv shows/movies/video games to play, so taking the time is okay. A sensor driven pooh collector might miss a lot of old dry poop, while a dumb collector isn't just going to get the poop, but also the buckeyes, walnuts and little sticks. Maybe a dead sparrow or two, but your yard is probably safer to mow without those things in it...
still, it's a fun thought experiment, isn't it?

I tried this a while back while working on a dog poop sensing roomba attachment.  Here are a couple of videos showing the concept...  The results weren't promising.

Part 1 (Methane Sensor)


Part 2 (Hydrogen Sulfide Sensor)


Tell you what, Damo2-if Jax

starts selling a dookie-sensor in his store, I'll buy a set and add it to this.  Otherwise, it would probably be considered "cruelty to robots" to try to make such a device and I'd be first against the wall when Skynet or Arcos comes on line ;-)

You've taken my shoutbox conversation to another level! I'm trying to make a sort of fluorescent emission spectometer sensor along the lines of a color detector sensor. You're thinking of a absorbtion-type spectrometer which is probably beyond our financial and my intellectual grasp here. Probably more the latter than the former. Nonetheless, I will consider it :)