Ideas to gage water in a rice paddy?
Tracy Allen
Posts: 6,664
I've been asked to come up with a method to figure out how much water there is in a rice paddy just before planting. This has to do with voracious snails that eat the shoots, and biocontrol using a soapy substance extracted from quinoa hulls. The pesticide is expensive, and the dose per volume needs to be well controlled. Each paddy covers about 4 hectares (200m x 200m), and the water is about 4 to 7 cm deep. The way they do it now is to have a couple of people walk around in the paddy with a meter stick, and while one takes a reading, the other writes it down in a notebook for later transcription to a computer for the calculation. They take several samples at the corners and at the center. The accuracy is not good.
One way to semi-automate this would be to instrument the meter stick with a depth sensor, a GPS, a data logger and a pushbutton. However, I think they are looking for something more sophisticated. The thing that comes to mind is a raft with depth sensor/GPS/wireless/data logger that can either be controlled from shore, or be preprogrammed to hit waypoints, lots of them.
How would one propel or build such a raft? Also, for the depth sensor, the coolest thing would probably be ultrasound, but at that shallow depth it would have to be an interferometer rather than a ping-type device. Or maybe could a servo controlled feeler.
Ideas?
One way to semi-automate this would be to instrument the meter stick with a depth sensor, a GPS, a data logger and a pushbutton. However, I think they are looking for something more sophisticated. The thing that comes to mind is a raft with depth sensor/GPS/wireless/data logger that can either be controlled from shore, or be preprogrammed to hit waypoints, lots of them.
How would one propel or build such a raft? Also, for the depth sensor, the coolest thing would probably be ultrasound, but at that shallow depth it would have to be an interferometer rather than a ping-type device. Or maybe could a servo controlled feeler.
Ideas?
Comments
Would the raft have to be automated to avoid running over the rice shoots?
At that small depth range, a laser line generator aimed at an angle and coupled with a linear array sensor might suffice to measure the bottom contour, so long as turbidity is not an issue.
-Phil
Of course better than a raft would be a quadcopter buzzing around lowering probes. It might look like a giant dragonfly laying eggs.
Back to the raft. I was trying to think or some sort of touch sensor that would work in the mud but I didn't really like most of the ideas I could come up with. One idea I kind of like is to use a weight with an accelerometer inside and have the ground detected by an abrupt acceleration. As I think about this a bit more I'm sure there are other low tech ways to detect the bottom. The tension on the line lowering the weight could be monitored in order to find the bottom. A hinged feeler gauge device might work.
If you had a rod on slight tension pointing downward, you could monitor the angle of the rod to determine the depth. I've often thought a magnetic encoder would make a good water proof knob of sorts. The magnetic encoder could be attached to the rod's hinge point to monitor the angle of the rod. There's sort of an example of a magnetic encoder used as a knob in the video in this post. An optical encoder on the hinge would be another option.
I agree it would be nice to have a non-contact method to detect the depth but I'd only be guessing (more than before) on what might work.
Interesting idea about the line generator. So, the instrument would measure the length of the projected line, is that right? Via a video camera?
Very interesting challenge and could be a lot of fun to solve! Here are the first thoughts that ran through my mind when I read your description:
- Google image search has several homemade airboats that might work better for rice paddies. I searched for "RC Airboat" and "RC Foam Airboat".
- Extreme idea: Quadcopter equipped with a "floating buoy" dangling from a line below the quadcopter. Fly to a position, drop to where the buoy is on the water, and the buoy contains electronics for the measurement (feeler, sonar, etc)
Sensing method could easily dictate the transport mechanism, but with a foam airboat, weight should not be a major concern. Another thought is a weighted lever that hangs off the end of the boat. Use an encoder to determine it's angle and based upon it's drop, the angle would vary. Use a servo with a spool/string to pull the lever back up. Here's a crude sketch:EDIT: It would appear that Duane and I think a lot a like, but I type slower............
"giant dragonfly laying eggs" hehe!
If there is a feeler on a post, it should have a blunt wide end. I'm told that the substrate is pretty hard, but that does vary over the area.
I think the raft could be made so it only draws a couple of centimeters, with a small propeller (screw - the processor would be on the raft :-)
The whole thing, including the smarts could be powered by a small LIPO battery. If people can walk around in there, the bottom must be fairly firm, although I would think it would be somewhat muddy if the shoots are to be planted in it. Also assume the shoots are planted with some space between rows, etc., so the craft would have to be programmed to follow the rows. (PING sensors to keep it centered in the rows? Or just whiskers?)
I'll bet there are any number of small RC boats that could be adapted to this...EDIT - Like they said. :blank:
It's a bit eerie how much we do (even the quadcopter). You have better pictures though.
Here's a diagram: http://forums.parallax.com/showthread.php/113366?p=813695&viewfull=1#post813695
-Phil
I like the idea of the air boat. About steering, the commercial "Alligator" says "Twin air rudders with adjustable single link between steering servo and rudders for fast, powerful response". I'm not sure how that works. I'll have to find time to follow the links for DIY airboats. You're right, there are a lot! How about a quad-copter props turned up to horizontal?
The probe does not have to be long. A linear gear of the sort that is used to raise/lower the radio aerial on a car might work, with a touch switch for contact with the bottom.
'
-Bob
Edit: if it sounds crazy, google "rice planter" or "rice transplanter". They're wheeled vehicles. The paddy looks more like a mud puddle than a lake.
I'd say 5 nodes as "...They take several samples at the corners and at the center..."; plus eTape sensors positioned not vertically but rather at certain angle to cover their whole range of 12" (as the max water level is expected to be below 10cm).
'
I think your way over thinking this.
The height can be predetermined by measuring the skirt height under cushion power motor. Thrust motor would have no consequence on that measurement.
Jim
.... thats a workable plan but will certainly present some challenging hurdles.
Edit- but in all honesty I think ANYTHING with a hovercraft is friggin cool and MUST be part of the solution! For uh, errr... engineerings' sake. It's just gotta be done!
Got a picture of the paddy in question?
One sensor per rice paddy's enough of course - I just tried to repeat "their approach" as close as possible
-Phil
Unless it's a US government operation.
On prerequisite for this technique is that the horizontal feature size of any depth anomalies would have to be much greater than the bot's wheel spacing.
-Phil
Yep, at least 101 ways to do the measurement. A plastic tube makes a good guide/container for a float. Ping pong balls make great floats. Add a strip of perf board and a photo interrupter to the float and you can measure depth to 0.1". Drill the holes out to 50 thou and you can get 0.05". Of course you may have to reduce the current to the led or paint the PCB black, but saving power is always a good thing.
I presume this application is for the 3rd world, so I suggest the solution below.
The answer lies in those bare-basic paddy tractors they use in SE Asia. (I love those things; just a chassis, portable diesel engine, a few belts and great big metal wheels. When they are finished plowing the paddy they disconnect the portable diesel engine and hook it up to their boat, food mill, electric generator or whatever needs power next!)
Make a robot with big spoked wheels that looks like like a paddy tractor. Use some of the spokes as depth gauges by fitting them with sliding floats with magnets and reed switches on the spokes.
Fit the paddy tractor robot with a GPS and data logging module.
Get a young lad to drive the robot up and down the paddy field using a standard hobby remote controller.
Download the GPS and depth data from the data capture into a software application and calculate your volumes.
Love the simplicity of this idea but how do you determine which spoke to take the measurement from?
I should add that although each paddy is around 4 hectares, there are hundreds of paddies, thousands of hectares. Thus a search for a quicker way to profile. The depth is not consistent across the paddy, and will change from one planting to the next. That is why a single point or small set of fixed points per paddy is not enough. Later on, once the profile is known, the fact that water seeks its own level can allow a single point measurement to track changes in total volume. How many points are necessary? I guess that is one thing that needs to be determined. The rice robot would presumably be deployed one paddy at a time.
I've been told now that the water will be quite clear, not a mud puddle at this stage. But I suppose that condition will change when a person or a vehicle or a robot or planting machinery stir it up. Photo? Here is one that conveys a sense of the enormity.
I like the ideas about the wheeled robot with an angular or linear encoder attached to a float or bottom probe. Something like an oversize boe-bot. It has a physical definiteness to it. I understand about the stilling issue. The 'bot could stop at points to allow ripples to settle.
The paddy tractor with a sensor on a spoke or spokes--mechanically simple. I suppose you would sync the measurement to the spoke position, acquire when pointed down.
Then there is that dragonfly. The copter with a dip stick.
The current project is not third world. It is in the EU, where the snails are invaders not yet established. I'm pretty sure they are looking for something with more panache than some poor guy's leg!
I suspect that the density of the sampling is critical to getting better accuracy. So a simple automated approach that does the whole field would achieve a lot more than just doing the corners and the center of the field.
And I suspect that keeping a historic record of depths should be useful as the contour of the underlying bed is unlikely to change much over time. To prove that is true, ask the farmers if they have figured out there own best estimate of what amount of chemical works. They may have arrived at some good figures via trial and error.
So the solution may be more in the maths than in deriving a fancy method or device.
Of course, you could consider installing survey benchmarks at the corner of every field and try to lock in a precise reference grid.
Anyway, the above suggestion by Bob (a couple of other posts have suggested something similar) is one which appeals to my gut feeling about what would work.
-Tor
If one could fly over with a quad-copter, photograph in the right combo of filter and spectrum, and then get a computer to determine average depth by color variation -- the whole process would be less expensive and more efficient. I just have not idea about the feasibility of such an approach. Some form of radar might be another good alternative.