Let's Make Robots!

High Frequency beacon

Hi guys

I'm working on a lawn mower. I'd like it to precisely know its position in order  to optimise its travel.  As it is working outside and on a more than 500 square meters, the IR beacon (needs to remain in sight) , the GPS(not sufficient precision) won't fi my needs. Any idea like HF beacon ? 

Comment viewing options

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

HI

I'm back for a short while not to say I'm not working any more  on my mower (the grass is still growing even in winter) but I'm spending all my free time on designing the main board. I'm using KICAD and freerouter. Both are very good and complete but need a long time for a beginner.

I hope I will be able to show the result before spring and fast growing grass!

As I said before, I've decided to take the option of the magnetic sensor and optical sensors on wheels.

 

I would possibly try something like this:

The robot has a fairly high power US transmitter, and you put receivers on three corners of your yard. The difference in receive time from the robots ping should allow you to estimate the position. Probably through some matrix equation for finding the closest approxmiate solution. (I forget what it's called) This will of course have to be RF'ed to the robot.

Possible extensions to this include:

Adding a fourth receiver way out of the plane of the other three, which should allow positioning in 3-space. Three receivers will give two possible solutions - one above the plane and one below. This may not be a problem if you always know which solution to pick - if say you have mounted the receivers on small towers to ensure "line-of-hearing", the robot will always be below.

Adding more receivers will cut down on error, and may allow the receivers to be self-calibrated - ping one of them in turn, and with same method vaguely outlined above, figure their relative positions.


I don't know the power requirements of doing something like this with ultrasound on this scale, and I haven't actualle solved the equations for it, but it "feels" computable.

 

edit: fudge, the math seems alot more involved than i had hoped for...

 Well, as an RF link is already required, the ping could be initiated remotely. That would make the times absolute, and the prolbem simple - none of this finding, the intersection point of circles with unknown radii. Actually the intersections of several hyperbolas. I still think it should be solvable, but definately not a linear problem.

The poster has not been on for 2 months.
Heh, my how time flies - what happened to November I ask?
Maybe hmn44 is just moping about the grass not growing in november. Now his mower has got nothing to do. Real robotocists would buy Astroturf!

Many thanks for all your comments ! I think we 've already evaluated many solutions and I think that the best outdoor solution would be a differential GPS that is : 1 GPS on the mower, 1 GPS on the base with a known reference position. The base station is then able to compute the tiny difference between the known position and the position given by its GPS. Then this difference signal is sent by radio to the mower whitch is then able to compute its own very precise position .

Very smart solution and reliable but I think a bit too much for my mower ! So I've decided to choose the compass and the underground wire.

 

This is probably too small for what your're doing, but this is what I'm looking to base my beacon nav system around.

http://norrislabs.com/Projects/WatchBot/index.html

http://norrislabs.com/

 

As far as knowing where I'm at, I'd be looking at a bit of trig to find that out based on 2 ref points and a known heading(compass bearing). As long as you have the beacons at set distances, you should be able to calculate where you are.

since my plan was to use it more as way points, I won't be going into the detail you may need.

Since ir isn't going to give you super range, why not make it a grid with intermidiate beacons?  transmit id's for each beacon so you know which is which and create a (hash)table or array that contains the list of all beacons so you know where each should be. 

This would give you quite a bit of what you need and you'd be able to pinpoint your location.

This looks to be a cool project btw..I'd love to see how things go with it....   :)

 

 

 

I found this site today when googling for "robot beacon". The site is by an active group in a Luxembourg school. It sports the Lego NXT throughout.

The page I'm linking here proposes to combine Ultra sound emitters that fire simultaniously around the target area. The triggering is by one of the beacons that sends an Infra red broadcast. The robot receives the both the light and the sounds and deducts its relative position based on the time differences between the received signals. Therefor this system utilizes the Time-of-Flight of sound through the air for location determanation.

A few advantages:
system does not need a beacon to be "reflecting" the bots broadcast
the bot does not need to look in the direction of the beacon
geometry is relatively mild on the MCU

The maker has developed his ideas all the way. The idea, the mechanics, the electronics (complete with pcb layouts in cad software) and programming (assembler PIC). That might be an advantage.  If you're willing to follow his footsteps exactly. I am favouring an adaptation to personal preferences. If there exist any in your lab.

As stated in my blog today, I have an interest in outdoors navigation. But as it is today, I am nowhere near starting a project like this. Anxious to learn about your adventures Mike!

8ik

I agree to the solution of the compass and the encoders on the wheels : the 2 things that are missing on my prototype !

 

Maybe the cheapest way would be the i2c comapass, a few underground wires, and encoders on the wheels. One problem might be the ground being uneven and the encoders/wheels loose traction.