Mount a strobe to the mower. Keep track of the mower/strobe with a video camera (fixed position) on the roof, and integrate that with a grid-coordinate map to guide/navigate against.
Another idea which I saw in a hospital once is use rfid. U can use the cheap unsecured tags bc theyll be buried in the ground. simply put them in a track a few feet apart in a track you want the mower to follow. Not only is it cheap itl be easy too bc rfid tags are alot easier to bury than wires. The parallax rfid reader wont have enough range on it...I think its in the range of like 8cm or somthing. http://www.amazon.com/LONG-RANGE-125KHZ-RS232-READER/dp/B003OFOO4Q You could buy somthing like this and set it up on a phidgets reader or if your really up for it mod the parallax reader and put a custom antenna on it. If your lucky you can get a parallax rfid reader from radioslack for 8 bucks on clearance I just picked up a few last week.
Thank you all for the really great input. First, my yard is large, about 1 acre, and is an odd shape. There are some obsticles like small trees, but they will not create any shade. I have been thinking about the IR beacons and my thoughts are that it will be 12V or 24V that will be charged by a solar cell/small battery system, easy install. I also thought that I can develope a program that will read each of the corners (or turning points) as a combonation, like I will place the mower, push a button to take position readings from all 4 beacons. Then move to the next corner, push botton to take more positoin readings. I will do this for all turning points to collect all the data. Then, having all this data, I will then place ths data into memory and use these point as reference points for programming turns. That is my current plan. After the first parimeter cut, I will then use cut grass for navigation.
Looking into the RFID tags, its my understaning that they can only be read from a few inches. If true, then they will have to be very very close to each other. Using this will be expensive if you have to around an acre will it not? If I am wrong, then please explain.
There is no fencing in my yard, but there are some surrounding bushes. Land is not level, has a very slight slope.
Well They can normaly only be read froma few inches away. However with antennas, lower frequency systems and descent tags id bet you could bury them at least 5-6 inches and still read them. A good book on this to get u started is "rfid toys" http://www.amazon.com/RFID-Toys-Projects-Entertainment-ExtremeTech/dp/B002QGSXRI/ref=sr_1_1?ie=UTF8&qid=1313416163&sr=8-1
The book has a really good intro to rfid and bs2 projects as well. I picked my copy up at ollies for 6 bucks about a month ago. If ur quick you may still get a copy cheap!
IMHO you're up for a lot of work with the IR beacons. Here is the rundown of things you have to solve first:
1. The transmitter (beacon) strength. Assuming your yard is a square you're looking at more than 200 feet on each side. If you plan on putting the beacon in the corners it may be impossible to find a strong enough IR diode.
2. The receiver sensitivity. It needs to be very sensitive because of 1. Therefore even small change in lighting conditions will effect it. The clouds or time of day etc.
3. You need at least 3 bacon hence 3 separate devices to worry about
4. Recognizing each beacon's signature. Try to use your DVD player and TV remote at the same time. None will work. Therefore you'd need to turn each beacon in a round robin fashion (they'd need to communicate with each other) or have a rotating sensor on your mower. You'd need to reduce the angle on the sensor too.
5. Since it's optical it'd be very easy to disrupt the location calculation simply by an object that's in path (pet or a person).
6. Complex position calculation algorithm.
Here is the RF fence idea:
1. The cable in the ground will have only one control point
2. Not sensitive to light changes
3. Not sensitive to "normal' obstacles (unless it’s a large metal object)
4. No need to detect signature because there is only 1 “transmitter”
5. The “receiver” is simple – actually you need 2 on both sides of the mower.
6. The search algorithm would be very simple. Here it is:
You put the mower anywhere in the yard. It drives straight first until it finds the signal. It turns left (can turn right too) and follows the cable making sure that the cable is between the 2 sensors. When the signal is gone it turns left and search for the cable again (the corner). Once it’s done with all 4 corners it will continue with the uncut grass as you pointed out before.
Unless your objective is to work on a cool idea that's high risk go for the beacons. If, on the other hand, you'd like a lot lower risk solution I'd recommend the RF fence.
While there are ways of dealing with the problems alex123 lists I am also beginning to agree that an RF wire around the perimeter may be the best approach. The hardware and control software would be simpler.
Well simple is all relative. You can buy IR beacons, and receivers. You can also buy RFID cards and receivers. Try buying a wire system. If you skip the dog, you can get the wire fence, and the collars, but the receivers, the collars give a non consistent pulse, or you have to design something. EVERYONE IS NOT A DESIGNER no matter how easy it is to some. So now that I made a full circle with all the ideas, does anyone have anymore?
Yianie,
The beacons will actually require lot more design work as I pointed out before. The RFIDs are really short range devices. You'd need to bury them every 20 inches or so in the ground. I wouldn't go that route unless I had unlimited funds.
Don't buy the pet invisible fence as it'd be too much work to adapt it to your needs. The wire can be anything you want. It can be the Cat5 ethernet cable. Just use one wire out of the 8 in it. Measure its resistance when you have 800ft of it. Use the Ohm's law to calculate what kind of drive strength you need for it. Roughly the copper wire has about 0.15ohm/m so you're looking at 40Ohm. At 12V that's 300mA. You're good with the driver IC I had the link to before.
As we talked about it before, you could use the relay coils for your detectors + $8 amp board. You need to add a diode in series and a 10uF cap to ground at the output of the amp. This will be your detector. Make 2 of them.
In all cases you need some kind of brain to run your algorithm and control the servos. You can’t avoid a micro anyway.
That's my take on it. Kwinn, correct me if I'm missing something here...
Alex, any possiblity you can send me a wiring diagram? Email Address: John.Vasos@gmail.com It would take me out of the tourment that I'm in.
Thank you,
John
I would consider that one of the best favors someone as ever done. Please know that no one seems to sell anything like this. NONE of the robot parts or microcontroller manufactures have anything like this. I will be forever in your debt. Please include both ends, wire, radio frequency (if that is what is used) parts, and receiver parts.
Thank you,
John
I think it'll be easier for me to just point you in the right direction rather than giving you the whole thing. Too much work...
1kHz generator
Use this http://www.ti.com/lit/ds/symlink/ne555.pdf - see page 10 figure 12. Calculate the RA, RB and C so the circuit generates a frequency close 1kHz. Use 12V for power and make sure you have a nice decoupling cap on 12V to GND close to the chip.
Detectors:
Take apart 2 small relays. The coils will be your detectors.
Amplifier
It took me awhile to find a single chip solution. The IN118 seems nice - only a few external parts. It looks like it may need a dual voltage power supply (or 2 batteries on the mower) or an inverter. I'd use a dual voltage bench top PS for the initial setup/testing. http://www.ti.com/lit/ds/symlink/ina118.pdf (The first circuit on Figure 3 gives you what we want). Again don't forget about the decoupling caps.
Setup
First use an oscilloscope to measure the output of the INA118 amp. Power up the loop driver and move the coil close to the wire. You should see some sine wave on the screen. Make sure you don't touch the circuit otherwise you'll get all sorts of things on the screen. At this point you have to tweak your amp gain by changing the RG resistor. Make sure the signal is strong ~ 10V pk-pk. Once you have that set up, you can add the diode and the capacitor. If this will feed a digital input pin I'd recommend an optocoupler to isolate the digital parts from the analog circuitry. Here is what I have in mind:
The cap value selection and the resistor calculations are next.
Sorry for the delay in responding. Busy dealing with another emergency that is taking a lot of time.
@alex123, the only thing missing as far as I can see is an adc to measure the voltage from the coils, although that could be built in to the micro.
Once the signal detected by the sensing coils exceeds a threshold level the difference between the levels could be used for steering.
The cat5 cable used could be shorter if you stripped out the individual wires and connected them together. The wire ends would need to be soldered together, silicone sealant placed on the uninsulated joints, and heat shrink tubing slid over the silicone seal and shrunk in place. Probably better to us a pair if you do this so you have a spare.
... the only thing missing as far as I can see is an adc to measure the voltage from the coils, although that could be built in to the micro.
Yup, could use an ADC. It'd be harder to optically isolate the analog signal from the opamp but still doable. We'd need some circuitry to limit the max voltage on the ADC input b/c the opamp can go close to its +V rail. A resistor + zener should suffice I guess.
My 'digital' idea may cause the mower to zigzag when it will be following the cable.
Alex
Here is the update with the ADC idea. Yeah, click on it.
I'm not sure if you know of not, the thumbnail diagram can be enlarged by clicking on it. I usually right-click and choose "Open Link in a New Tab."
Alex and Kwinn,
Thanks for all this great information. I'm following along with interest.
Yianie,
Is your yard enclosed? I've often wondered about some sort of robotic lawn mower but one (of many) factors keeping my from trying it is safety concerns. I'm afraid a robotic lawn mower would be just too interesting for kids to leave alone. I'm afraid I'd fail to safeguard against all the ways a child could injure themselves with it.
Can 1KHz be detected easily from a distance? I needed to find a buried wire a few years ago that was part of my yard sprinkler system. I found that frequencies around 500 KHz worked well.
I also thought about building a robotic lawn mower. One method I considered was to put optical markers at the corners of the lot and using a digital imaging device to detect the markers. The position and orientation of the lawn mower could be determined through triangulation of two markers. This is similar to the IR idea.
The other method I considered was to build a local "GPS" system in my yard. My "GPS satellites" would be located at various points around my yard. The distance to each "satellite" would be determined by measuring the phase and/or time of the received signal. I think this would be allowed within the HAM radio bands and a HAM license, but I haven't explored whether this could be done.
Can 1KHz be detected easily from a distance? I needed to find a buried wire a few years ago that was part of my yard sprinkler system. I found that frequencies around 500 KHz worked well.
We don't need distance. The coils should be as close to the ground as possible. No more than 10 inches above the ground. I picked 1khz as it's a nice round number. If higher frequencies would work better that's fine. Obviously certain frequencies will give you a better electromagnetic field range. I was afraid that we may get more of the higher order effects (due to the parasitic capacitance and inductance of the loop) with higher frequencies. That would manifest in the driver chip getting too hot and eventually failing.
This is the fun part anyway, right?
I was afraid that we may get more of the higher order effects (due to the parasitic capacitance and inductance of the loop) with higher frequencies
Frequency doesn't matter. It's the rise time and fall times that matter. That's what I meant anyway...
Here is the loop driver schematic. We need a series resistor on the output of the driver to limit the current drawn on voltage transitions. This resistor may also be used to detect if the loop gets broken. That may become handly sometimes. A friend of mine gave me this cool idea.
The yard is not enclosed. Some bushes but also some areas not enclosed at all. Thank you for your diagrams. Still beyond my understanding and soldering skills. Was hoping for something more off the shelf. Long distance card reader.
It doesn't get much simpler than that. All ICs are in DIP packages... I played with this type of parts when I was a kid (30+ years ago). Believe me, it's not that complicated... Buy a proto board and give it a shot...
You know when someone says "add this" , "and adjust that" do not expect some new to electronics to experiment. Yes, a wiring diagram, is just that,a wiring diagram. Yes, I am the one that needs a complete, point A to point B wiring diagram. You are very kind to go through all this work and yes I will admit to my own stupidity, but yes, I do need a complete wiring diagram. Thank you again, Yiannie.
You know when someone says "add this" , "and adjust that" do not expect some new to electronics to experiment. Yes, a wiring diagram, is just that,a wiring diagram. Yes, I am the one that needs a complete, point A to point B wiring diagram. You are very kind to go through all this work and yes I will admit to my own stupidity, but yes, I do need a complete wiring diagram. Thank you again, Yiannie.
Yianie, I have to tell you that some experimenting will be needed to make this work. The basic ideas are relatively simple, and the technology they are based on have been in commercial use for a number of years, but unless you are willing to spend a substantial amount of money for commercial "off the shelf" parts you will have to do the experimenting. Once you have decided to proceed with this project the folks on this forum will provide a lot of help, but you will have to do the actual work. Think of it as a great learning experience. I will help as much as possible but right now I am very busy so I may not have that much time. I am sure alex123 will do the same.
Very well said. I am not expressing any disappointment other than my own fustration. I am very knowledgable with many things, but electronics is something of a huge challange. I konw I have to experiment to get things right. I, for a breif moment I forgot how lucky I, and many others are to have kind people like all of you and this forum to come to for help. I may not get it now, but I will soon. Thank you all for your help.
Hey ive been following this from the beginning now. When you finaly get a working prototyp going post up a vid of it working. Ive always wanted to see this sort of system in action.
Comments
I think it is do-able with some additional hardware and work. First though can you answer the questions below.
1 - How will you power the beacons? Battery or a line powered supply (line powered is preferable).
2 - If the answer to 1 is battery what voltage will you be using (12 - 24V preferred)?
3 - Is your yard fenced with something that blocs light or does it have plants at the corners that blocks some of the ambient light?
http://www.amazon.com/LONG-RANGE-125KHZ-RS232-READER/dp/B003OFOO4Q You could buy somthing like this and set it up on a phidgets reader or if your really up for it mod the parallax reader and put a custom antenna on it. If your lucky you can get a parallax rfid reader from radioslack for 8 bucks on clearance I just picked up a few last week.
Looking into the RFID tags, its my understaning that they can only be read from a few inches. If true, then they will have to be very very close to each other. Using this will be expensive if you have to around an acre will it not? If I am wrong, then please explain.
There is no fencing in my yard, but there are some surrounding bushes. Land is not level, has a very slight slope.
The book has a really good intro to rfid and bs2 projects as well. I picked my copy up at ollies for 6 bucks about a month ago. If ur quick you may still get a copy cheap!
1. The transmitter (beacon) strength. Assuming your yard is a square you're looking at more than 200 feet on each side. If you plan on putting the beacon in the corners it may be impossible to find a strong enough IR diode.
2. The receiver sensitivity. It needs to be very sensitive because of 1. Therefore even small change in lighting conditions will effect it. The clouds or time of day etc.
3. You need at least 3 bacon hence 3 separate devices to worry about
4. Recognizing each beacon's signature. Try to use your DVD player and TV remote at the same time. None will work. Therefore you'd need to turn each beacon in a round robin fashion (they'd need to communicate with each other) or have a rotating sensor on your mower. You'd need to reduce the angle on the sensor too.
5. Since it's optical it'd be very easy to disrupt the location calculation simply by an object that's in path (pet or a person).
6. Complex position calculation algorithm.
Here is the RF fence idea:
1. The cable in the ground will have only one control point
2. Not sensitive to light changes
3. Not sensitive to "normal' obstacles (unless it’s a large metal object)
4. No need to detect signature because there is only 1 “transmitter”
5. The “receiver” is simple – actually you need 2 on both sides of the mower.
6. The search algorithm would be very simple. Here it is:
You put the mower anywhere in the yard. It drives straight first until it finds the signal. It turns left (can turn right too) and follows the cable making sure that the cable is between the 2 sensors. When the signal is gone it turns left and search for the cable again (the corner). Once it’s done with all 4 corners it will continue with the uncut grass as you pointed out before.
Unless your objective is to work on a cool idea that's high risk go for the beacons. If, on the other hand, you'd like a lot lower risk solution I'd recommend the RF fence.
P.S. I ate the goat.......
The beacons will actually require lot more design work as I pointed out before. The RFIDs are really short range devices. You'd need to bury them every 20 inches or so in the ground. I wouldn't go that route unless I had unlimited funds.
Don't buy the pet invisible fence as it'd be too much work to adapt it to your needs. The wire can be anything you want. It can be the Cat5 ethernet cable. Just use one wire out of the 8 in it. Measure its resistance when you have 800ft of it. Use the Ohm's law to calculate what kind of drive strength you need for it. Roughly the copper wire has about 0.15ohm/m so you're looking at 40Ohm. At 12V that's 300mA. You're good with the driver IC I had the link to before.
As we talked about it before, you could use the relay coils for your detectors + $8 amp board. You need to add a diode in series and a 10uF cap to ground at the output of the amp. This will be your detector. Make 2 of them.
In all cases you need some kind of brain to run your algorithm and control the servos. You can’t avoid a micro anyway.
That's my take on it. Kwinn, correct me if I'm missing something here...
Thank you,
John
I've often wondered how a wire following robot works.
Duane
Thank you,
John
1kHz generator
Use this http://www.ti.com/lit/ds/symlink/ne555.pdf - see page 10 figure 12. Calculate the RA, RB and C so the circuit generates a frequency close 1kHz. Use 12V for power and make sure you have a nice decoupling cap on 12V to GND close to the chip.
Loop driver
Use the part from the parallax website:
http://www.parallax.com/Store/Components/AllIntegratedCircuits/tabid/154/CategoryID/31/List/0/SortField/0/catpageindex/3/Level/a/ProductID/608/Default.aspx
Connect all 4 drivers together in parallel. Pins 2-7-10-15 (this is your input) and 3-6-11-15 (this is your output). Connect all enables together and connected them to VCC1. Connect VCC1 and VCC2 to 12V power supply and GND pins to ground. Connect all inputs to the 555 output (pin 3). Connect all outputs to one end of the cable, ground the other end on the board.
Detectors:
Take apart 2 small relays. The coils will be your detectors.
Amplifier
It took me awhile to find a single chip solution. The IN118 seems nice - only a few external parts. It looks like it may need a dual voltage power supply (or 2 batteries on the mower) or an inverter. I'd use a dual voltage bench top PS for the initial setup/testing.
http://www.ti.com/lit/ds/symlink/ina118.pdf (The first circuit on Figure 3 gives you what we want). Again don't forget about the decoupling caps.
Setup
First use an oscilloscope to measure the output of the INA118 amp. Power up the loop driver and move the coil close to the wire. You should see some sine wave on the screen. Make sure you don't touch the circuit otherwise you'll get all sorts of things on the screen. At this point you have to tweak your amp gain by changing the RG resistor. Make sure the signal is strong ~ 10V pk-pk. Once you have that set up, you can add the diode and the capacitor. If this will feed a digital input pin I'd recommend an optocoupler to isolate the digital parts from the analog circuitry. Here is what I have in mind:
The cap value selection and the resistor calculations are next.
Comments are welcome.
@alex123, the only thing missing as far as I can see is an adc to measure the voltage from the coils, although that could be built in to the micro.
Once the signal detected by the sensing coils exceeds a threshold level the difference between the levels could be used for steering.
The cat5 cable used could be shorter if you stripped out the individual wires and connected them together. The wire ends would need to be soldered together, silicone sealant placed on the uninsulated joints, and heat shrink tubing slid over the silicone seal and shrunk in place. Probably better to us a pair if you do this so you have a spare.
Yup, could use an ADC. It'd be harder to optically isolate the analog signal from the opamp but still doable. We'd need some circuitry to limit the max voltage on the ADC input b/c the opamp can go close to its +V rail. A resistor + zener should suffice I guess.
My 'digital' idea may cause the mower to zigzag when it will be following the cable.
Alex
Here is the update with the ADC idea. Yeah, click on it.
I'm not sure if you know of not, the thumbnail diagram can be enlarged by clicking on it. I usually right-click and choose "Open Link in a New Tab."
Alex and Kwinn,
Thanks for all this great information. I'm following along with interest.
Yianie,
Is your yard enclosed? I've often wondered about some sort of robotic lawn mower but one (of many) factors keeping my from trying it is safety concerns. I'm afraid a robotic lawn mower would be just too interesting for kids to leave alone. I'm afraid I'd fail to safeguard against all the ways a child could injure themselves with it.
Duane
I also thought about building a robotic lawn mower. One method I considered was to put optical markers at the corners of the lot and using a digital imaging device to detect the markers. The position and orientation of the lawn mower could be determined through triangulation of two markers. This is similar to the IR idea.
The other method I considered was to build a local "GPS" system in my yard. My "GPS satellites" would be located at various points around my yard. The distance to each "satellite" would be determined by measuring the phase and/or time of the received signal. I think this would be allowed within the HAM radio bands and a HAM license, but I haven't explored whether this could be done.
We don't need distance. The coils should be as close to the ground as possible. No more than 10 inches above the ground. I picked 1khz as it's a nice round number. If higher frequencies would work better that's fine. Obviously certain frequencies will give you a better electromagnetic field range. I was afraid that we may get more of the higher order effects (due to the parasitic capacitance and inductance of the loop) with higher frequencies. That would manifest in the driver chip getting too hot and eventually failing.
This is the fun part anyway, right?
Here is the loop driver schematic. We need a series resistor on the output of the driver to limit the current drawn on voltage transitions. This resistor may also be used to detect if the loop gets broken. That may become handly sometimes. A friend of mine gave me this cool idea.
Yianie, I have to tell you that some experimenting will be needed to make this work. The basic ideas are relatively simple, and the technology they are based on have been in commercial use for a number of years, but unless you are willing to spend a substantial amount of money for commercial "off the shelf" parts you will have to do the experimenting. Once you have decided to proceed with this project the folks on this forum will provide a lot of help, but you will have to do the actual work. Think of it as a great learning experience. I will help as much as possible but right now I am very busy so I may not have that much time. I am sure alex123 will do the same.
1kHz generator
Loop driver
Detectors (self explanitory)
Amplifier
Thank you for the additional help. I am going to build it, just would like some additional insite.