Latest El Reg project: Rise of the Robot Sheep
After seven years of faithful service, my lawn-clipping droid Mowbot has had to be retired... and a replacement is hard to obtain. Rather than face an ever-growing lawn, I've decided it's time to unlock the inventing shed and seek some reader advice. Robots were supposed to be doing pretty much all our manual labour by this …
This topic is closed for new posts.
Electrolux where are you
I have had a Electrolux Automower for 5 years. While the iPhone gets out a new version every year and gets lots of interest, this stuff doesn't change. I have been ready to buy an update for the last 2 years but are still using the old one. The first and really simple upgrade would be the necessity of guide wires. The machine already measures how far it goes by counting wheel rotation plus it has programmed direction changes so it know the angles of direction changes. Even without a GPS it should easily construct a map of the lawn, and with todays flash prices keep up to date where it is and have been (sharable by wifi). Second and equally simple upgrade is to get rid of need to reprogram timing between spring and autumn. The machine already measures necessary force on cutter so it should easily be able to find out where grass is growing fastest and visit accordingly and rest when it is not necessary to cut. Thirdly a camera with color analysis should be able to tell where and what kind of fertlizer is missing. Fouth the machine should be able to deliver this fertilizer. Fifth doing the lawn edges would give time to drink even more beer on Saturdays. Electrolux, I am ready any day to test a beta.
PORTMAN
Probably Overengineered Robotic Turf Mowing Autonomous Navigator
Freaking lasers mounted on the head of a squirrel is what I want, and I want it now.
An Idea for the mowing
I'm thinking a cylinder wrapped around a rotini-like spiraling blade.
Start with a sheet metal tube, cutting a section out of the tube to make it more of a "C" shape. Line the tube with chicken wire and cover the gap (grass gets in, but not toys, kitties, etc). Have the spiral blades fairly close to the CW. Lining the tube with the chicken wire with a close tolerance for the blades will mulch the grass, so no need to worry about bins, composting, or such.
By putting the tube/grate opening slightly forward of bottom-dead-center, the tube would also bend the grass, making the coveted lines (or words). Or put the tube on a servo, moving the opening fore/aft/center and adjust the whole cutting assembly up and down on the fly and you could program it to do patterns. A happy face, I think, to let all know I am glad I didn't have to mow this.
The obvious name is Kusanagi, but I have to start cooking now, so I cannot pick a backronym.
Paradigm shift
How about instead of treating lawn mowing as a once-per-week or so activity, treat it as a constant thing. As soon as a blade of grass gets to a certain length, zap the top off with a laser. Constant scanning as a specific height means one or so shots per minute, with some software to avoid shooting larger objects such as children.
Arduino(s) FTW.
DGPS should be fine right up to the point where the 'merkins find out somebody has oil and decide the people should be freed from the shackles of their dictatorial regime, hammering the accuracy of the GPS signal on the day they invade (yeah, I know they said they'd not degrade the accuracy again but if they have to pick a fight with someone their own size who knows)
Seriously though, bump sensors and/or ultrasonics for detecting small objects on the grass, a gimballed scanning laser on top with retro reflectors in corners of the grass, a map and a bit of maths should be a fairly good way of working out where you are.
Plus it'd look uber cool on a misty day if you chose a visible laser wavelength.
dreams
when I was a child 30+ years ago, I had a book about the future, jet packs with head up displays, bountiful energy from nuclear fusion, instead we got the solar powered parking meter.
Solar powered parking meters
They're really just solar powered? Awesome. Think of the fun I could have with a thick black texta or a can of black spray paint... ;)
Can I program
it to go on a killing rampage if it gets bored of going up and down your lawn
I would do that to the robots at work, but the boss, rather unsportingly, will fire me for it....
Wouldn't it be easier to simply start with a lawnmower and fit a propulsion and guidance system?
I mean, no need to completely re-invent the wheel completely. I'd very much like to see the first person with a working model though.
All the lawnmowers I've seen recently actually have a propulsion system (a small electrical motor driving the rear wheels, usually not sufficient to move the thing but easing the work somewhat).
The problem with that plan is that the whole thing will have to be modified so much that it is probably easier to start almost from scratch (of course existing motor blocks can and should be used; that's outside of the gardenshed-boffinry limits). Fitting a propulsion system on something that was not designed to accept one is going to be a major problem in itself; then there is the direction; then you need to get rid of the dead-man security system; etc...
Unless you want to start from one of these assisted-direction little tractor-like things? That would be an easy job (I believe all that would be needed would be to crack the control system for the assisted direction and throttle, and plug it into a control server) but that's going to be a huge machine, and hugely expensive, too.
Obvious choice
The base should be an Oldsmobile Delta Royal 88, modified to run on steam, with big spinning blades at the front.
Also, there is a typo in the article. You typed "avoiding children" when you surely meant "chasing".
More seriously, chasing high grass in real time sounds fun but it is probably simpler to enter the lawn's geometry beforehand and calculate the optimal path (step-wise, should be a doodlewith the right optical encoders), with a sensor to detect obstacles (with path re-calculation of course). Whatever system you use to decide of the path, keeping track of the position by the optical encoder steps, not by triangulation as suggested higher up, is probably the way to go. Of course you will need a calibration step first (so that the control system knows the dimensions of the area in terms of encoder steps), but it will make the programming much easier (basically logo-turtle-like). I know, it's no fun, but it is simple, and in software "smart and fancy" often means "likely to fail". Plus, that will make writing "tasteful" messages (or stripes, for the unimaginative) very easy.
I don't have any suggestion as to which kind of encoder/microcontroller combination would be best, as the one I am working with, although reliable and durable, won't last long in a vibrating, dust-saturated environment. You need some heavy-duty stuff. I wonder what they use in the metallurgical industry...
StephenC
The stripes are produced by the roller - not by the cylindrical cutter. I have a rotary mower that produces stripes this way.
Yes, the use of a cylindrical blade is normally to cut very close to the ground. It's usually reserved for the kind of mowing associated with bowling or putting greens.
Much as I'd like a lawn that good I think a normal rotating blade and roller will do the job. Although it will also need a strimmer attachment for the edges...
think small, think cluster
If I can make a small suggestion. If you are going to do all the hard work off the robot, make small and multiple. If the robots are cheap and dumb, then they are cheap. Bigger lawn more mini bots. If someone should take one, you can much easier to replace then one expensive one.
Compost Bin?
For picking up the grass and unloading into the compost bin, I would imagine a leaf vac/blower would do the job if you fit a valve. Have an outflow tube a bit like a combine harvester but using air pressure.
Also would be good to use something like an android smartphone or tablet... something with a USB port so the can connect to a suitable sensor array.... can just undock it then to tweak the software.
Turf Uniform Reduction Networked Equipment - Reel-type
I would suggest a small reel (perhaps 3" dia 6" long) with guard wires parallel to the rotation and no capture of clippings...basically a Mad Max Dirt Devil. If you make the reel small enough, it won't need enough power to sever toes and whatnot and the guard would let the grass through (and even guide it for cutting) without allowing curious paws to get any more than a manicure.
Or you could put your whirling death machine inside a mesh ball.
Some thoughts
How about buying one of those weed cutters that has a plastic wire spinning at high speed? It should cut grass just fine but probably just convince the owner of any fingers put in it's path that they (the fingers that is) would be better off somewhere else. If you bought a wireless one you would get both a battery and a charger for the lawnmower as well.
As for guidance, how about having the mower emit an infrared pulse every few seconds. You would then have tree strategically placed recievers that would look for their own pulse (one, two or three short pulses, possibly after an initialization pulse to make sure they are looking). The recievers would then emit an ultrasound pulse that the mower would hear. It would then calculate the distance to each reciever. Poor man's gps. It should then be a simple matter to program the path the mower should use.
I agree with previous posts that the arduino would be a perfect fit for this project
Stripes
The stripes come from the rear roller, NOT from cylindrical blades!
Most (petrol driven, user holds handle at the rear) mowers also use the rear roller for drive because it is heavy and has a large surface area (compared to the front wheels) and therefore traction.
Why not try...
Guinea Pigs. Forget goats, sheep, rabbits etc. Guinea Pigs will give a nice even cut and fertilise at the same time.
Cylindrical? Get reel!
The stripes are left by the roller on the mower, not the blades. My 3 circular blades, 6' total width, are followed by a 6' roller, so I can leave stripes that can be seen from space, but striping is for stupid suburbanites with too much time on their hands. (And I can mow most such yards with one pass up and one pass back, so it's not very well striped.) Meanwhile, I'm composting the clippings back into the yard, thickening the sod.
Reel mowers require a LOT of torque, and are very finicky about being razor-sharp and closely-adjusted against the bed knife. I recommend a flail mower segmented finely, so each section can closely follow the ground, but spinning swords require the least power.
Incidentally, goats have the largest stomach-area/weight ration of any grazer and don't like anything close to the ground, probably an evolutionary adaptation to parasites, so they eat the roses before the grass. And, having had hundreds of ovines, you can only eat a lamb if it has no name. Once the kids name it, it's a pet, not lunch, and you can't house-break sheep. (At least they only leave pellets, about the size of rabbit berries, but they do piss bright yellow.)
"...I'm composting the clippings back into the yard, thickening the sod."
Allowing the clippings to remain on the lawn creates what's referred to in gardening circles as "thatch". This causes waterlogging of the surface, which destroys the root mat and kills the lawn.
Compost the clippings elsewhere and dig the results into the flower beds where it'll do some good. Best thing for the lawn is to go over it with a fork, making rows of holes, followed by a light dusting of sand. This then washes into the holes made by the fork tines and improves the flow of moisture through the surface, preventing waterlogging of the root mat.
No, I do not know how I know that.......
a bot by any other name...
in keeping with the reg ideal of naming spb items after media friendly ladies I humbly suggest:
Generally
Autonomous
Garden
Assistant
Stripes
The blades don't do the stripes, the roller does, so a rotary cutter won't help.
You need a heavy roller to flatten the grass to give that stately home effect. Or a light roller pushed down with a load of lead acid batteries, perhaps.
eco friendly ...
what about using some of that high water psi drilling kit to cut the grass; then it gets watered as you work
Recycling
"...but we'll not be recycling any other parts of the old chap any more than we'll be building a chassis out of granddad's femurs. That also means anyone else who wants to have a crack can use our designs without having to buy a Mowbot first..."
That also means...granddad's femurs are still available for your other projects.
El Reg Project Naming Convention
How about -
IMOGEN - Independent Mowing Operation Garden Environment Navigator -
There is a 'turk' (south wales expression) 'bint' who meets the criteria of previously honoured ladies of ill repute.
of sheep and mowers
I was thinking a reel mower would be safest until I read Dan Paul's idea for a hedge clipper. Either could work and you'd want to go with whichever is easiest to keep at the right height given your lawn's micro-topography and the mower's wheels/suspension. Everyone knows that lawns need less fertilizer if you leave the clippings on them rather than bag and remove them, so that should be your goal. Reel mowers work best if you mow frequently and that would be best for mulching. If you find yourself building up thatch, you'll have to invent a robot aerator/de-thatcher, which could be a lot of fun in and of itself. As to guidance, my first thought was to use a system that triangulates using those mirror balls people put in their gardens. If the robot had a bright light on it, the spherical mirror would reflect it at a very precise point, no matter where the robot was. Use three different color balls and lenses with three different filters to allow the robot to distinguish among them. If the cameras were at the top of the robot pointed up and there were rotating 45 degree mirrors mounted above them, then the camera would get a flash whenever the mirror pointed at its ball. The bearing to that ball could be read from the rotor on the mirror and a computer should be able quickly to determine the mowers location every full mirror rotation. I believe that early IR seeking anti-aircraft missiles used similar devices.
Get hold of an old flymo
Cause the blades are metal, sharpenable, potentially lethal, and very effective. And they cut in any direction. Might even have an old spare one somewhere...
The cylinder doesn't give you the stripes, and cause it relies on having a second blade to pass against, it one directional and takes more effort to spin.
It's the roller that gives you the stripe, and you'll need to work out a turning technique so not too much of the stripe is lost at the tops and bottoms. But of course rollers will take way more energy to move around.
Mulching is ok, saves collecting stuff, and what about a feeder/seeder to add to it for that really lustrous green lawn?
You can always add a flame job to the paint if it's too lame for you...
How adverse would you be to having the controls electric, but the power from a petrol engine? A little moped engine would give you more than enough grunt to run a 6V circuit too?
Brain of the beast
I think the path calculations should be done locally (i.e. hardwired, not on a remote server as suggested somewhere in the article). Remote control is just asking for problems. What if you lose the link? There can be a remote oversight, but the core of the control should be local. Also, for safety reasons, if contact with the remote control channel is lost the mower needs a local system so that it can park itself safely, say, on the nearest Iranian desert*.
I suggest using a Ben Nanonote as the local brain. It's cheap, relatively inexpensive (100 bucks), resilient (I dropped mine a few times while in the tube; it takes 1.5 m falls on concrete -with forward momentum, even- without a complaint), fully copyleft, flashable at will, it has ethernet over usb or can be fitted with a WiFi microSD card (if you can find one). 336 MHz MIPS-compatible chip, 2GB NAND, 32 MB RAM, with a small colour monitor and a full-featured keyboard for in-situ interaction if needed; what's not to like?
http://en.qi-hardware.com/wiki/Hardware-Ben
There's plenty of space to put your -legally obtained- music tracks, so attach a 50 quids sound system from Costco -with a massive subwoofer- to the lawn-mowing core and presto!, you got yourself an efficient neighbour-angerer.
The current shipped image is an OpenWRT with Python2.6, Lisp, 4th, gforth, perl I think, etc (and of course C). Even Octave, the open-source MatLab clone! so it could be used out of the box; of course it's all open source, so the sky (and your cross-compile toolchain) is the limit.
If you keep the microSD slot for wireless comm, the only possible channel to control the mowbot would be the USB port of the Ben (I'm guessing ethernet over usb would be the most convenient way to talk to whatever board is talking to the microcontrollers - and encoders, but what do I know? see disclaimer below). I can submit a tentative circuit board design if needed**. I do know my way around optical encoders, microcontrollers and such, and high-level languages are not a problem, but I have litterally zero experience in linking both through USB, so YMMV, as the cool kids say. The datastream needed is really more a datatrickle, so the Ben's USB 2.0 will be plenty fast. That's a good start, right? RIGHT?
*Nah, just kidding
** Caveat lector: all circuit board designs I might submit will be heavily inspired by pre-existing, proprietary designs belonging to various companies, all of which are long dead -but some might have been bought by Oracle, some of the angles might be rounded, which I hear is patented by Apple, and component colour might involve brown, so beware the Zune patent pool; your status = warned. The designs predate the "patent the obvious" goldrush, which is good, but almost always involve a 386 chip, which is bad. Well, for some values of "bad".
Patch
:%s/cheap, relatively inexpensive/relatively inexpensive/g
# Obviously.
side benefits
There is a nethack port for OpenWRT on the very Ben NanoNote, so in theory it could verily be used to draw your very own nethack levels. On your very lawn. How very cool is that?
(this message sponsored by VerY Inc., proud owner of the "very" trademark. VerY Inc., now a part of Oracle. Expect some rebranding as soon as you're used to the price hike)
Go ahead with a big cylindrical reel mower. Start with a 36 inch wide snow blower with tracks, change the blades, up the gearing and power it with an old air cooled VW engine. It will sure keep the lawn tidy if the second stage is up to tossing everything, even the neighbors cat, into the neighbors yard.
Brazilian ?
No no no, that would mow the lawn clean all over, apart from a thin strip up the middle
Some ideas
How about buying one of those weed cutters that has a plastic wire spinning at high speed? It should cut grass just fine but probably just convince the owner of any fingers put in it's path that they (the fingers that is) would be better off somewhere else. If you bought a wireless one you would get both a battery and a charger for the lawnmower as well.
As for guidance, how about having the mower emit an infrared pulse every few seconds. You would then have tree strategically placed recievers that would look for their own pulse (one, two or three short pulses, possibly after an initialization pulse to make sure they are looking). The recievers would then emit an ultrasound pulse that the mower would hear. It would then calculate the distance to each reciever. Poor man's gps. It should then be a simple matter to program the path the mower should use.
I agree with previous posts that the arduino would be a perfect fit for this project
One frankly disappointing aspect of the suggestions seems to be the focus on health and safety above all. Daleks et al were not known for being slow underpowered things that have to return to the charging dock frequently, ok so they did take out a few fleshies too but hey. Let’s consider some user requirements that would make the design transferable to a wide number of readers – such as ability to cope with longer grass, slopes, uneven lawns, molehills, fallen leaves, no outdoor electricity supply, and how about a cost ceiling too?
Combine with HOVERCRAFT!
Blades - design for cutting and LIFT!
Just like the old electric 'FlyMo' things, except with on board energy (electric, IC or nuke, your choice) and completely autonomous.
I think you could get the control gear cheap from Iran.
Semi-autonomous Hovering Electric-Engine-Propeller Flying Under Computed-Kinetic-Error Restriction
Fuel?
All this talk of binning the cut crass, rather than dump it in the compost bin surely we could utilize the composting process on board to produce heat and therefore power?
Over to the Bi-chemists for the technical stuff.
Other outdoor devices, to get ideas from
Farmboz isn't actually a robot, but has quite a solid mechanical design: http://eirepreneur.blogs.com/eirepreneur/2011/01/irish-made-wonder-robot-herds-sheep-mows-lawn.html and http://clikire.snappages.com/farmboz-3.htm
Slugbot is an autonomous self-feeding garden predator: http://www.wired.com/gadgets/miscellaneous/news/2001/10/47156?currentPage=all and http://www.ias.uwe.ac.uk/Robots/slugbot.htm
I am thinking something along the lines of Terrafirminator as featured in Gnomeo and Juliet...
Re-consider Sheep
You know, you might reconsider the sheep. You could use sheep to do the mowing and build the robot to pick up their poo. You know, something with a dustpan and broom on the front...it'd be cute!
Rise of the Robot Sheep?
Tsk, tsk, tsk.
Any fule kno that the correct term is *Electric* Sheep.....
Cylindrical cutter
If you use a separate motor to run the cutting bits, torque to propel the thing around your lawn becomes less of a concern Of course, you still have to deal with the rolling resistance, especially on a sloped lawn, but that's well within the capabilities of a pair of wheelchair motors. Traction of the drive wheels on (possibly wet) grass would be more of a problem.
Cutting blades
Why not use plastic ones? I don't mean strimmer wire, I mean plastic blades as supplied for various electric mowers. I use an old Flymo from time to time which has two plastic cutters about 3 inches long. At operating speed they cut very effectively, they eventually break but are cheap to replace and if you hit anything especially solid (rock, free roaming tortoise, fallen over elderly relative..) they disintegrate without causing any visits to A&E.
Getting a good cut seems to require speed more than power. Oh and the lawn needs to be not too wet, a moisture sensor might be good to include.
Weight distribution and augmented guinea pigs
Type of mower strikes me as important. If you want a roller to make stripes I guess you'll need some weight over that roller, so engine or battery packs over or behind the roller? (mid-engined or rear engined like a 911 sir?) If however you want a hover motor which has its advantages (moves more freely, look vaguely cool, blows stuff all over the place frightening bystanders) then weight will be a problem. I don't know if they do hover mowers that can do stripes though the old flymo I use does have a roller.
Or you could just use augmented guinea pigs. Strap a lawn cutting jacket to them complete with battery pack, head mounted camera, smart phone brain (might want to avoid using one of the new 4"+ house brick ones) and maybe a frikkin laser beam just because. Then spend your time trying to prevent the neighbours seeing your guinea pig cyborg minions and shopping you to the RSPCA.
This topic is closed for new posts.
