Need advice on robotic mower project

[brentonw2004]

Hello everyone! Recently I have been thinking about trying to do another upgrade to my lawnmower. Currently I have installed two wheelchair motors, battery system, etc. to a normal gas powered motor to make it a remote controlled mower. I have been doing some thinking on how I could make it completely autonomous for fairly simple yards. I realize that there is a system where you burry a wire in the ground and it can find its boundaries through the wire, but I would like to try to make an easier installation method for it. Does anyone have an opinion as to how I can get the mower to know its position in the yard in some type of coordinate system. I had thought about doing this once before by using something like GPS, but I have yet to find a module with enough accuracy (<.1m). Do they make such a module or is there a way of achieving such accuracy? I know that scientists use extremely high accuracy GPS in measuring changing fault lines in California. How do they do this? If anyone know of any good GPS tutorial sites, or just any good resources as to where to buy GPS modules fairly cheap, it would be helpful. The other idea that I had is to maybe place something like two radio beakons in the yard and have each of them measure the distance to a beakon on the mower, then triangulate the mowers position relative to the beakons. Does anyone know how accurate a system like this could be, and how to design one? Thanks in advance for any advice!!

 

[Someone Electro]

GPS is anything but cheap

 

[tansis]

Couple of years ago DARPA was tinkering with battlefield robots that referenced thier position using low freaquency sound waves.


Technical paper for 3-D Position Sensing Using Ultrasonics
**broken link removed**

Sound Localization Algorithm on a Robot
**broken link removed**

 

[stevez]

A good friend of mine (fellow engineer) works part time on a large, modern farm. He was telling me about planting crops then cultivating or treating with the aid of a system that incorporates GPS and some local benchmarks for the precision needed. I understood that the system would automatically steer the equipment based on info on where seeds were planted. I'm sure that's way beyond what you want but researching agricutural uses of GPS may help with your project.

I was thinking that lasers could triangulate position and follow the mower but that seems rather complicated. You have at least two, maybe three stationary laser units that would scan for the mower and the mower would have a receiver that could differentiate between each stationary unit. The stationary units would report direction only - then a computer would have to calculate the position.

 

[Roboticinfo]

Hello everyone! Recently I have been thinking about trying to do another upgrade to my lawnmower. Currently I have installed two wheelchair motors, battery system, etc. to a normal gas powered motor to make it a remote controlled mower. I have been doing some thinking on how I could make it completely autonomous for fairly simple yards. I realize that there is a system where you burry a wire in the ground and it can find its boundaries through the wire, but I would like to try to make an easier installation method for it. Does anyone have an opinion as to how I can get the mower to know its position in the yard in some type of coordinate system. I had thought about doing this once before by using something like GPS, but I have yet to find a module with enough accuracy (<.1m). Do they make such a module or is there a way of achieving such accuracy? I know that scientists use extremely high accuracy GPS in measuring changing fault lines in California. How do they do this? If anyone know of any good GPS tutorial sites, or just any good resources as to where to buy GPS modules fairly cheap, it would be helpful. The other idea that I had is to maybe place something like two radio beakons in the yard and have each of them measure the distance to a beakon on the mower, then triangulate the mowers position relative to the beakons. Does anyone know how accurate a system like this could be, and how to design one? Thanks in advance for any advice!!

You can use an encoder and use a timer two turn the motor left and right at certain times.
The only problem is that the drive motors will slip and you will end up short if you decied to use an encoder.

 

[Oznog]

GPS becomes quite accurate when correction signals are included via differential GPS.

The error is GPS signals is quite repeatable in a particular area and time. If you place a receiver on a stake at a known location and take its GPS reading,
you can calculate the error vs its known location.

If you broadcast the error to a nearby receiver via radio, it becomes extremely accurate, likely within inches. Most of the position error happens because the individual satellites drift a few feet either way and the atmospheric conditions delay the signal inconsistently due to changes in its diffraction constant. All this will be the same between receivers and slow to change.

There is also a network of reference stations that broadcast the error correction over a fairly wide area. Many GPS modules have an input for those signals if you have a radio receiver. It's probably not going to be as accurate as doing your own without many miles of separation between the reference and your receiver.

 

[audioguru]

You can use an encoder and use a timer two turn the motor left and right at certain times.
The only problem is that the drive motors will slip and you will end up short if you decied to use an encoder.

Like NASA's Mars Rovers that are still running around on Mars nearly 1.5 years longer than expected. They slip going uphill, and the engineers on earth don't know about it until the telemetry gets here about 8 hours later.
One wore out a tire and got stuck in sand for a few weeks but is out now. https://marsrovers.jpl.nasa.gov/home/http://marsrovers.jpl.nasa.gov/home/

 

[chiba]

Just an idea, as it's already remote controlled, have you considered controlling your MowBot from your computer? It would be pretty easy to hook a transmitter up to a printer port? to me it would seem easier to let the pc do all the number crunching and controlling and leave the mowBot to just cut. The next step would be getting feedback from the bot or controling it to follow a simple map or even a manual control with a camera attached, that way you can drive it into the back shed or chase the cat around the back yard

 

[tansis]

Counting wheel revolutions to work out the bots position is a tried an tested method that works very well, that is until a loss of traction incident occurs. The same system is used in the humble and very common ball mouse and we all know how stubborn they can become when the ball fails to make good contact with the internal rollers. Now what if a sensor or two working in a similiar fashion as an optical mouse were employed on the underside of the robot looking directly at the ground then even if the wheels slip the position information would still be accurate. Such as system could be implemented with commercially available ultrasonics
**broken link removed**

 

[chiba]

Now what if a sensor or two working in a similiar fashion as an optical mouse were employed on the underside of the robot looking directly at the ground then even if the wheels slip the position information would still be accurate

How would it deal with the uneven surfaces such as un-mown grass? My optical mouse doesn't seem to want to work even when it's 1/4" from the surface, i believe they have use a focus point, an uneven surface may mess everything up?
How about if we put a censor on two wheels, that way if one spins at different rate to the other then slippage could be calculated and accounted for or simply put it on the non-driven wheel?

 

[tansis]

Using an optical sensor presents all kinds of problems in this particular application, which is why I suggested ultrasonics though it could just as easily be done with microwaves from a gunn diode. The key operating priciple is that of doppler shift and an uneven surface actually helps the process.

--------------------------


I wonder if the differential gearbox design of the "South Pointing Chariot" could be of some use :lol:

 

[Oznog]

Friendly Robotics' Robomower, the most popular robotic mower out there (I have one!) uses a magnetic compass sensor to detect heading. It has a wheel tachometer too. No GPS, it detects a loop of staked wire as well as bumper sensors. It cannot operate outside the loop except by manual control.

 

[plot]

A good friend of mine (fellow engineer) works part time on a large, modern farm. He was telling me about planting crops then cultivating or treating with the aid of a system that incorporates GPS and some local benchmarks for the precision needed. I understood that the system would automatically steer the equipment based on info on where seeds were planted. I'm sure that's way beyond what you want but researching agricutural uses of GPS may help with your project.

I was thinking that lasers could triangulate position and follow the mower but that seems rather complicated. You have at least two, maybe three stationary laser units that would scan for the mower and the mower would have a receiver that could differentiate between each stationary unit. The stationary units would report direction only - then a computer would have to calculate the position.

John Deere sells GPS systems for there tractors, basically put in the route and it'll do whatever to the field, no driver needed. So... there GPS systems must be pretty accurate.

for the lawnmower, maybe it could just sense the height of the grass... if it's over 3 inches, mow that part, under 2 inches, it's already been mowed, or something? I guess that would be a little complex though.

 

[Roboticinfo]

A good friend of mine (fellow engineer) works part time on a large, modern farm. He was telling me about planting crops then cultivating or treating with the aid of a system that incorporates GPS and some local benchmarks for the precision needed. I understood that the system would automatically steer the equipment based on info on where seeds were planted. I'm sure that's way beyond what you want but researching agricutural uses of GPS may help with your project.

I was thinking that lasers could triangulate position and follow the mower but that seems rather complicated. You have at least two, maybe three stationary laser units that would scan for the mower and the mower would have a receiver that could differentiate between each stationary unit. The stationary units would report direction only - then a computer would have to calculate the position.

John Deere sells GPS systems for there tractors, basically put in the route and it'll do whatever to the field, no driver needed. So... there GPS systems must be pretty accurate.

for the lawnmower, maybe it could just sense the height of the grass... if it's over 3 inches, mow that part, under 2 inches, it's already been mowed, or something? I guess that would be a little complex though.

Actually plot that is a very good idea, I would of never thought of that.
for that you could use a IR sensor mounted somewhere that is leved and basicly mow anything that hits the IR.
Not a good idea if you have kids or cats or dogs, but would work.

 

[DigiTan]

I had thought about doing this once before by using something like GPS, but I have yet to find a module with enough accuracy (<.1m).

What exactly are your cost and time limits? If you can purchase two GPS receivers and survey the exact location for your main GPS antenna, you will be able to send error corrections to your mower--maybe with one of those low-cost RF modules. Basically the correction signal would be a 3D vector telling your mower to adjust its "preceived" longitude, lattitude, and altitude.

In fact, this is the exact basis for DGPS, except since your differential system would be contained to your yard, it's likey you can beat the 10m resolution limit of a modern (D)GPS.

If you're still interested, you can try to track down a low-cost unit like a Trimble "SV6" for as low as $18. This is a typical surplus price, but shouldn't (have to) pay more than $40 for one. I have one (unused) unit, and a Ham buddy of mine says it will generate everything from location to time signals to numbers of satellites visible. Plus it works with both active and passive GPS antennas. Anyway, it's a little elaborate, but if time is no limit, it's worth looking into.

 

[tansis]

How about a line follower ?

Spray paint a line around the perimeter of the area to be mowed then let the bot use "bump-n-go" logic. If a method could be found to detect the difference between the hieght of cut and uncut leaves of grass this would speed up the process.


Thinking of spray paint on grass...a roving robotic painter that could transfer the image of choice to the grass...or any other suitable flat surface...