TL;DR: IR LEDs and phototransistors are good only for a small range of angles and need to be pointed at each other to really work. I need something less 'directional' that can work in a range of 3-15cm while providing a rsolution of nearly 5mm or better.
Hi. I need to track the position of a moving point. For this, I plan to compare its distances from two stationary points and do some math to calculate its coordinates.
All I need are the ratios a/c and b/c. Here X is the target and A and B are the two stationary points. Point Y is another stationary point that acts as a reference for the distances. All the distances are very likely going to be between 3 cm and 15 cm.
I am currently considering IR LEDs and phototransistors. I can use the LED as the target X and the phototransistors as A and B. The LED will emit light which, depending on the distance, will vary in intensity. This will be picked up by the phototransistors. I can compare the readings in A and B to figure out the ratio. The problem is that they only work in a very small range of angles and pretty much need to be pointed at each other. This severely restricts the movement of target X. I need something less 'directional' (Maybe like a light bulb). Is IR a good idea at all? What sensor/arrangement would be ideal for it?
IR seems like a decent option. I would use 2 LED on X pointing to A and B. Only 1 LED should be open at a time. You should get a strong beam on one fototransistor and a weak one on the other. After that you would need to experiment and implement your own algorithm. There are LED with high directivity ( usually dependent on the LED shape)
Another option would be using ultrasonic sensors. You use one transducer (can transmit and receive) on A and one on B. You transmit from A and listen to A and B. Do the same from B. Depending on the responses you can figure out the position. The ultrasonic sensors have 1 mm - 3mm resolution and the low frequency ones have high directivity. This won't work if the object X is very small/ sharp since ultrasonic waves won't bounce. You should be able to do 10 measurements/s. You can can go higher only of there are no objects in nearby vicinity.
CalinTamaian:
IR seems like a decent option. I would use 2 LED on X pointing to A and B. Only 1 LED should be open at a time. You should get a strong beam on one fototransistor and a weak one on the other. After that you would need to experiment and implement your own algorithm. There are LED with high directivity ( usually dependent on the LED shape)
Another option would be using ultrasonic sensors. You use one transducer (can transmit and receive) on A and one on B. You transmit from A and listen to A and B. Do the same from B. Depending on the responses you can figure out the position. The ultrasonic sensors have 1 mm - 3mm resolution and the low frequency ones have high directivity. This won't work if the object X is very small/ sharp since ultrasonic waves won't bounce. You should be able to do 10 measurements/s. You can can go higher only of there are no objects in nearby vicinity.
Thanks for the reply! The two-LED option will probably not work because X moves around a lot. Can I partially cover the outside of the LED with a reflective surface? Maybe that could redirect the light to spread equally in all directions.
Using ultrasonic transducers seems like a brilliant idea. Maybe I can attach a transducer to X and let A and B recieve the signal from it. What will be the effective range of the ultrasonic transducers? I have read that they are not as accurate for small distances.
Laser displacement sensor are designed for controlling moving equipment. They are used in a variety of expensive machines where some part of the machine needs to move to an exact position but very rapidly. A measurement is taken with the laser displacement sensor and then a command is sent to move the machine to exactly that position at full speed or some very fast speed. A crude analogy would be a laser targeting sensor for an artillery gun. Unlike missle tracking lasers where the missile follows the laser, a targeting range finder is a laser that is pointed at the enemy tank to get the range which is then keyed into the fire control computer which tells the tank what angle to adjust the barrel to prior to firing the round, and ballistics and gravity take care of the rest.
Is the moving object moving in a straight line or zig-zagging ?
What exactly are these references and the object? Without a description, I can't think of a generic method that works regardless what they are. Do you have a picture of your set up?
liudr:
What exactly are these references and the object? Without a description, I can't think of a generic method that works regardless what they are. Do you have a picture of your set up?
Thanks for the reply. Here are some (crappy) MS Paint illustrations:
I want to study different combinations of two spirographs (at different speeds, angles etc.). There is a rectangular surface/region with a sensor/tracker (shown in red) at each of its corners. The two gray lines are rods that extend to the two different spirographs which we want to combine. The spirographs will move the rods, and hence their intersection (the target point), in a crazy but mathematical pattern. I want to track and study this point. Side view:
Top view:
Technically, I only need 2 sensors/trackers. If I use 4, I can average the results from each pair to get more accuracy. I hope this clarifies it.
If IR LEDs and phototransistors had enough range and angle, I could use an LED (facing down) at the tracking point and phototransistor at the corners. If ultrasonic worked in small ranges (like 3-6 cm), I could stick an emitter/transducer to the tracking point, and sensors/transducers at the corners.
OK I understand a bit more now. Do you mean you have two toothed gears that move against each other and a pen tip trapped somewhere between the gears and records the graphs? If you had paper under it, would you get the graphs? Spirographs are the resultant curves, not the objects that are rolling around each other. And you can solve it or make parametric plots of what happens, simplest being a wheel rolling on flat surface.
If you want to track the pen tip, why not using a capacitive touch surface and metallic pen tip for the sensing? It's better than 5mm accuracy. You can also use an active digitizer tablet and its pen and track it like tracking any mouse movement on computer using Processing. Why devising something while technology already exists? Any reasons the above two solutions would fail? Again I still don't have complete understanding of your situation.
Any non-physically connected sensor will need a means to have the transmitter pointing at the reflector to keep the reflections perpendicular.
So for instance narrow beam LED or IR LED or ultrasonic and sensor mounted on a stepper motor that can scan back & forth & find the strongest (or only) reflection; use that angle, and the angle from another corner & do the math ("simple" triangulation) to calculate position.
For example, say the bottom left corner was "a" and the bottom right corner was "b", and the upper left/right were "c" and "d". If angle made by "a" and line "p" is known, and angle made by "b" and "q" is known, can do the math to find the intersecting point, which can be no further distance than a-c and a-d, or corner to corner.
This is before the capacitive touch screens became popular. I've seen a South Korean company making a product to turn regular screen into "touch" screen with a device and a pen. What they do is emitting IR as timing pulse and ultrasonic as distance ranging waves. The device probably has stereo microphone for ultrasonic waves that start listening as soon as the device sees IR flashes from the pen. The pen shall emit ultrasonic wave. You get a complete (x,y) solution only if you assume the device is always above the pen.
Using ultrasonic transducers seems like a brilliant idea. Maybe I can attach a transducer to X and let A and B recieve the signal from it. What will be the effective range of the ultrasonic transducers? I have read that they are not as accurate for small distances.
[/quote]
I guess it depends on the definition on "accurate" and "small distances". From what I've seen/tested so far ultrasonic sensors have between 1-3 mm resolution in 20 mm - 1000 mm. This depends mostly on the transducers and how good is the software on the PIC/8051 on the module. Above 1000mm you can expect 2-5 mm. Bellow 20 mm , most of them cannot measure. The modules are cheap so you can test it yourself:
Lasers have of course bigger resolution (0.1 - 2 mm) but they are more expensive, have low directivity and higher power consumption. If you want your project battery powered using lasers/LED...it can be tricky. Also, if you plan to use this setup outside or in place where sunlight is present then IR is not an option.
liudr:
OK I understand a bit more now. Do you mean you have two toothed gears that move against each other and a pen tip trapped somewhere between the gears and records the graphs? If you had paper under it, would you get the graphs? Spirographs are the resultant curves, not the objects that are rolling around each other. And you can solve it or make parametric plots of what happens, simplest being a wheel rolling on flat surface.
If you want to track the pen tip, why not using a capacitive touch surface and metallic pen tip for the sensing? It's better than 5mm accuracy. You can also use an active digitizer tablet and its pen and track it like tracking any mouse movement on computer using Processing. Why devising something while technology already exists? Any reasons the above two solutions would fail? Again I still don't have complete understanding of your situation.
Sorry for the confusion. I did mean the rolling toothed gears, and if I were to place a paper under it, I would get graphs. I intend to study the graphs generated by a combination of two such rolling-gear-apparatus-things.
Capcitive touch surface and digitizer tablets are great for both precision and accuracy, but they get really expensive as the size increases. My budget is of about $150, but since I am a college student, less project cost equates to more pizzas. Using IR LEDs or ultrasonic transducers would save me much more money.
CalinTamaian:
I guess it depends on the definition on "accurate" and "small distances". From what I've seen/tested so far ultrasonic sensors have between 1-3 mm resolution in 20 mm - 1000 mm. This depends mostly on the transducers and how good is the software on the PIC/8051 on the module. Above 1000mm you can expect 2-5 mm. Bellow 20 mm , most of them cannot measure. The modules are cheap so you can test it yourself:
Lasers have of course bigger resolution (0.1 - 2 mm) but they are more expensive, have low directivity and higher power consumption. If you want your project battery powered using lasers/LED...it can be tricky. Also, if you plan to use this setup outside or in place where sunlight is present then IR is not an option.
Ultrasonic sensors sound ideal for this project. One more question though: How much will the accuracy suffer if the medium of transmission of the waves is not air? For example, what if instead of pointing the emitter directly towards the sensor, I point it down at the surface?
Sound waves, instead of travelling directly through air, could travel through the surface (maybe like ripples in water). That would solve the problem of directivity.
I can conduct the experiment in a dark room, so IR won't be a problem either.
Ultrasonic sensors sound ideal for this project. One more question though: How much will the accuracy suffer if the medium of transmission of the waves is not air? For example, what if instead of pointing the emitter directly towards the sensor, I point it down at the surface?
No way. The ultrasonic sensors are based on sound reflection, not vibrations. So pointing the sensor to the ground will cause a 0 reading. You need to use either multiple sensors on the moving object , either one of every stationary point.Environment has its influence . Temperature,pressure and air humidity will affect the speed of sound and thus your measurements. But unless you are not using them in extreme changing conditions you should be fine (you can also make software compensation for temperature/air humidity(the sensors are cheap dirt).
Another option would be using a gyro/accelerometer/compass combo. You don't need the stationary points, only sensors on the moving object. You will always get x,y,z difference between the actual position and the initial point . You can check the MPU- 9150 device., for example. I don't have experience with them so cannot tell you the price/precision.
Ultrasonic sensors sound ideal for this project. One more question though: How much will the accuracy suffer if the medium of transmission of the waves is not air? For example, what if instead of pointing the emitter directly towards the sensor, I point it down at the surface?
No way. The ultrasonic sensors are based on sound reflection, not vibrations. So pointing the sensor to the ground will cause a 0 reading. You need to use either multiple sensors on the moving object , either one of every stationary point.Environment has its influence . Temperature,pressure and air humidity will affect the speed of sound and thus your measurements. But unless you are not using them in extreme changing conditions you should be fine (you can also make software compensation for temperature/air humidity(the sensors are cheap dirt).
Another option would be using a gyro/accelerometer/compass combo. You don't need the stationary points, only sensors on the moving object. You will always get x,y,z difference between the actual position and the initial point . You can check the MPU- 9150 device., for example. I don't have experience with them so cannot tell you the price/precision.
I am looking into it right now