Can tank treads offer precise movement?

Would the slipping action of the tread in these turns make keeping track of position (via revolution sensors on the drive motors/gearboxes) a near impossibility, or is it actually not that big a problem?

If the ground is uneven, you'll get inaccuracies in position due to shifting or sliding off small rocks or non-level ground.

Is it a big problem? I guess it depends on how rough the terrain is. If the vehicle climbs a small rock at the time it makes a turn, it will skew a couple of centimeters maybe. A bigger rock will cause a larger error.

If GPS is not an option, maybe a pair of radio transmitters will can be used to establish the vehicle's position via triangulation (or RDF-radio direction finding).

Tracked vehicles tend to churn up the ground as they turn. And as they have to skid turn, turning also consumes a lot of power. Why not use a multiwheel approach with drive on all the wheels?

"Why not use a multiwheel approach with drive on all the wheels?"

Two rows of wheels much like two tracks. All the same never mind. Some of the wheels will have to slide same as the tracks, being rubber compounds the problem unless all of the wheels can be made to individually track or steer about a vertical axis.

Your solution is to buy one of RC and to modify it, there are stores specialized in which you look for, like: RC Car

Many years ago I worked on a mobile pipe mill (several hundred tons) that had caterpillar tracks. We used electric motors on both tracks with 4096ppr tachs on the motor, giving very fine resolution on the tracks. However, we were going essentially straight ahead with only small angular changes. The actual rig then followed a laser survey beam and tracked along the beam.

For indoor applications it is quite common to track a cable imbedded in the floor. However I am not familiar with the details.

I would personally favour a multi wheel approach, similar to the Mars probe of a few years ago, it could climb over rocks!

Larger wheels climb better.....

It would appear that you would need some sort of highly accurate "navigation" system and even GPS has its limits.....I would guess that this would be the problem that needs solving first!!

If there are walls around the area concerned, how about laser measurement of the distance to the walls (at least two of them!) + orientation info (electronic compass?) to locate the device accurately.....laser measuring tapes have been around for years and are consistent and very accurate!!

As an alternative to tracks look at the self propelled modular transporter (SPMT) design, it is a wheeled design with two rows of twin axle boogies; each boogie can rotate 270 Degrees. The height of the boogies are controlled hydraulically, and the tyre foot print is very close to the overall plan of the entire unit, so the unit more or less contours and object when moving over it. SPMTS are used for heavy transportation, where they do not have to move over obstacles like rocks, but they do have to overcome sharp changes in gradient over short distances (within the SPMTs length) when off loading from a barge or crossing a bridging structure. The SPMT's have programmed steering patterns that allow them to turn 360 degrees on their own centre they have a true 2 degrees of freedom in their steering plane and do not require skidding or churning to alter directions. See the web sites below for the main manufacturers, there information is limited, so search Google for SPMT's to see them in action.

www.koegelkamag.com

www.scheuerle.de

"they have a true 2 degrees of freedom in their steering plane"

That should be 3 degrees of freedom it has yaw as well.

Hey, all... Thanks for all the great info. Guess I can go ahead and reveal what I'm building since most of the people who are likely to say nothing more than "That's already been done, go out and buy one!" have already read and passed by this article. I'm planning a Robotic lawn mower.

Obviously, the unit's not going to be doing any rock climbing, but still has to deal with the inaccuracies of a non-planar surface. My other concern with tank tracks would be that the ground might get dug up whenever the unit makes a turn! The obvious things I have to deal with are staying within a perimeter, avoiding boundaries within the perimeter (trees, signs, gardens) and making even, relatively parallel and slightly overlapping passes. Anyone up for a group project?

Playing fields or home lawns?

Anyone that can afford a robotic lawn mower can afford a lawn as flat as a golf green, so 6 to 8 inch steerable powered wheels should be OK.

Navigation will be the biggest problem, tracking a buried wire, or a visual system might work. Safety will be a big problem.

My guess is that wheels will suffice for this application. While adding tracks could improve mobility, it also increases weight and requires more power to move it. To put it in perspective, on a 50 ton tank, about 5 tons is simply the tracks. But even if you did use tracks/treads like a tank, the unit is probably not going to be heavy enough to chew up the ground (much).

I agree with the other posting from GW about safety and navigation. I can't wait to see the evening news when one of the mowers "escapes"!

Or the neighbours kid or miniature dog that gets mulched.

Hmm, risk management says no way. Not a big enough market to justify the liability.

Hmm, risk management says no way. Not a big enough market to justify the liability.

Who said anything about a market? I'm just making one of these for myself. Call me greedy if you want, I just call myself lazy.

Okay, back to being serious, one of the easiest ways I could think of to do this is just make it like one of those "line following" robots, and bury a trace wire like was suggested earlier, using a similar system to the "invisible fences" for dogs. Again, with the lazy bit, the idea of running a wire under my lawn, back and forth, getting the spacing just right, and all the rest that goes with it just seemed... inefficient. The next solution I came to was to use the invisible fence to make a perimeter boundary, or just program the boundary into the memory of the bot and let it go to town. The problem with that would be "islands" inside the boundary where the bot shouldn't go.

I have just shy of a couple hundred ideas on how to get it done, from the simple (define a boundary with the electric fence and let the thing bounce around inside it) to the complex (multiple Pan-Tilt security cams connected to and controlled by a central computer, which tracks the motion of the mower by triangulating it's position using the video and pan/tilt measurements from two different cameras (like NASCAR has been using for a couple years now) and controlling the mower wirelessly according to a pre-surveyed map of the yard-in-question.) Unfortunately, I'm only one man, and I have enough ideas to keep a whole team of people busy for years (And barely enough money for one project at a time)

I think the neighbours kid pushing the mower might be the answer.

Pay him $10. For $200 you have the grass cut for a year.

Tracks might be a good answer to lawns with a lot of moss amongst the grass, power driven wheeled mowers often need a good push when getting stuck in moss. A tracked mower may be a better design for the future of robotic mowers if they want to become widespread. My thinking is to keep the tracks short and still use a jockey wheel at the front, to give a tricycle set up exactly like the existing ones, the two tracks would be used for driving and steering, the jockey wheel just for support. For control I would think about a laser set up, as a boundary splitting it into line of sight sections. Not all lawns are as straight forward as a postage stamp and would be difficult to use GPS or a counter system, the idea with these robotic mowers is to leave them running to their own devises like a goat that won't eat your washing. You would need other sensors to internal objects within the boundary and a simple program to go around an obstacle.

It's been done but you want to do it differently. That's the spirit!

The early robotic lawn mowers had a grass length sensor which allowed it to sense the cut grass. When it sensed the short grass, it was programmed to turn right (or left, depending) and when it sensed the long grass, it turned left. This simple programming kept it at the cut grass' edge. Before you set it off, however, you had to pre-cut around the perimeter of the lawn so that it had a boundary to stay in. It's sort of like a blind man with a walking stick to tell him about the terrain.

'Seems like your planning to get your robot to "memorize" the shape and size of your lawn so that you can just lay it down and it'll cut the grass with no help from you.

Wheels or tracks should work fine. The main problem is getting the position information to the unit. I suggest a three antenna pulse transmitter.

You send out pulses from three antennas at regular intervals. The time difference of arrival will allow it to calculate its position within the lawn and even out of it.

Alot of people have suggested Antenna Pulse Triangulation, but for some reason I was under the impression that for small areas (like the average suburban lawn) the area between the antennae wasn't big enough to allow for the fine resolution needed to maintain the border between the grass and "Grandma's Award Winning Tulip Garden." Maybe I'm wrong. Maybe if I wasn't so lazy I'd actually run the numbers instead of just sitting here pestering you guys. :-P Actually, I've never really built that sort of circuit so I don't know how precise that sort of device can be, as far as time measurement goes. Again, The only thing standing between me and the information is laziness, so i should probably fix that and look it up...

What was I thinking? For radio-based triangulation to work, you'll need lawn about the size of a small city!

Okay, let's change that a bit. To use radio, you could use directional antennas. The bearing to the antennas with respect to the lawn mower gives you the position.

Alternatively, you could use ultrasonic sound pulses instead of radio.

I just saw a show on TV where this guy rigged his purely mechanical lawnmower to move along the edge of the cut grass. It would straddle the cut and uncut portion and move along by itself. He had to make a precut strip along the boundary as well as around the trees and shrubbery on his lawn. It worked fine! I didn't catch what year it was but it looked like it was the 1950's or 60's.

Alternatively, you could use ultrasonic sound pulses instead of radio.

Wouldn't I need to find a speaker capable of producing the sounds at over 120dB to compete with the lawnmower's engine? Thanks for the suggestions though, I'm getting a lot of great ideas, so keep 'em coming!

Wouldn't I need to find a speaker capable of producing the sounds at over 120dB to compete with the lawn mower's engine?

Not really, ultrasonic speakers and detectors are optimized for detecting ultrasound. They are practically deaf to sound within the audible range. Besides, the circuit can be set up so that it ignores audible sound. As for the power, that depends on the size of your lawn.

I still feel that lasers are the way to go, if sensing the precut grass is for you NOT an option...

You seems so lazy mate, I wonder if u'll make it... Laser is the best solution for your problem...

Cut the grass with lasers? Cool!

Neuter the cat at the same time .

What makes you think I can't neuter a cat with a regular, non-laser lawnmower?

"I just call myself lazy"

Didn't think you would have the running shoes.

Some problems there:-

you would have to have a completely flat lawn

Neutering the cat might be the least of your problems - like cutting your owners feet off for example.....

We were actually thinking more of lasers as a means of navigating around the lawn......

Sorry, just free associating. Off on another tangent.

Mind wandering around and being amused at ideas.

CO2 industrial lasers?

Use them for missile defence when not cutting the grass?

Did someone say Nark Nark at the door?

I am not complaining in anyway, we must always lighten up in whatever we are doing, its better for our blood pressure!

Have a great day.