navigation beacons

[arod]

I am entering a robotics competition where there are two beacons placed in two corners of a square playing field (shown as the "O" below in my ascii art lol).

-----------------------
|:::::::::::::::::::::::::|
|:::::::::::::::::::::::::|
|:::::::::::::::::::::::::|
|:::::::::::::::::::::::::|
|:::::::::::::::::::::::::|
|:::::::::::::::::::::::::|
|:::::::::::::::::::::::::|
|O::::::::::::::::::::::O|
-----------------------

The left beacon modulates its IR light at 2.5kHz and the right beacon modulates IR light at 5.0 kHz. I realize it might be possible to get positional information from these two beacons. If you are able to find the angle to each beacon relative to your robot, you could do some trig and get your position. My first question is how would one detect this modulated IR light (50% duty cycle). Would it be as simple as designing a circuit with a phototransistor/photodiode, a comparator for a threshold value, and connecting the output to a microcontroller? Are there readily available modules that will detect and be able to distinguish between the two signals of IR light at these frequencies? Every commercial module that I have found is specifically for a carrier frequency of 38-40kHz. This module could then be mounted to a servo of some type to get the angular position of the light. Is this even what the competition designers had in mind when they set up two separate beacons. Any advice or suggestions are welcome. I am pretty new to robotics, but I am fairly well versed in electronics.

 

[dknguyen]

I've never worked with IR, but I think if you just add a filter (I mean frequency-filter, but a wavelength filter might also be helpful) onto the circuit that you said it would work. I also think you would need a circular array and a way to pick out the brightest one- otherwise you would need to increase mechanical complexity by adding a pan unit which would greatly slow down response time. You might need a lenses (IR sensors with beuilt in focused lenses) for good directionality.

I also suppose you could directly measure the frequency of the square waves using an input compare module on a uC, though I am not sure how it would react to two square wave inputs going in at the same time and if you could separate them out. If you had lense I guess you could always assume the particular photo detector was only looking at one beacon and "round" the measured frequency to the closer frequency between the two beacons and assume it is that.

It would have been much preferred if the designers added a third beacons and the beacons had IR+ultrasonic pulse output

 

[jpanhalt]

Are the beacons parallel, as shown, or intersecting?

John

 

[dknguyen]

I assumed they were omnidirectional.

 

[jpanhalt]

I assumed they were omnidirectional.

They were not drawn that way, so I didn't assume it. My guess is that they are not omnidirectional, but rather directional with a fairly broad dispersion angle of say, 15 to 25 degrees, and intersecting.

Several months ago I suggested in response to one of the line follower-type questions that someone could try a little different approach modeled on the old beacon system for aircraft navigation (sometimes known as the the A&N airway system as A = dit, da and N = da, dit).

See these links:
http://www.navfltsm.addr.com/ndb-nav-history.htm
https://www.electro-tech-online.com/threads/need-information-about-ir-detector.24561/

If the beacons intersect and have different modulations, it would seem like a good opportunity to try that approach. John

 

[dknguyen]

WHat is telling you atht they are directional? I don't see it. Just looks like two mysterious round thing sitting at the corner of a square area.

 

[Nigel Goodwin]

A quick google found this:

**broken link removed**

It's not really what you're after, but I thought it was quite interesting.

 

[jpanhalt]

WHat is telling you atht they are directional? I don't see it. Just looks like two mysterious round thing sitting at the corner of a square area.

What was telling you they were omnidirectional apropos your earlier assumtion? So, I asked a simple question of the OP and have yet to receive the answer. In the absence of that informattion, I was simply making a suggestion. John

 

[dknguyen]

I was just curious as to the way you were reading the picture. I just figured that uncovered flashing photo emitters were cheaper and simpler than directional ones.

 

[jpanhalt]

It was the heavy, dark vertical lines that made me think directional. Moreover, single IRED have broad, but not omnidirectional , emission. John

 

[arod]

They are directional IR transmitters. Each beacon has 6 IR LED's that are wired in series (shown below). Sorry for not clarifying that in my initial post. The outer edge of my ascii picture just indicate the physical boundaries of the playing field (4 walls). Thanks for all the input/suggestions. **broken link removed**

 

[bobledoux]

A micro controller could count the period between light pulses and determine which beacon was being read.

Mount an IR receiver on a small stepper motor. Have it sweep a full circle. Establish a step on its sweep that is "straight ahead," from the robot perspective. Then rotate until the first beacon is read at its strongest strength level. Record that step number. Continue rotating until the second beacon is likewise seen. Record the step number. The difference in step numbers times the motor step angle is the angle between the beacons.

Here is an article with some pertinent math:

 

[jpanhalt]

If there were just one emitter on each side, it would seem that intensity measurements (as suggested in the A/N airway analogy) could be relatively easily calibrated. With multiple emitters on each side, I suspect the calibration would be considerably more complex. The beam divergance and intensity for each emitter is probably slightly different, and there has been no mention of how or whether the emitters are aimed.

Thus, I agree that in this scenario, a triangulation method as shown in bobledoux's link is a better approach. John

 

[Pommie]

Reading the link posted, it would appear that it is impossible to work out where you are with only 2 beacons.

Arod, do you have a link to the competition?

Mike.

 

[arod]

Reading the link posted, it would appear that it is impossible to work out where you are with only 2 beacons.

Arod, do you have a link to the competition?

**broken link removed**

You should be able to work out position with a little trig since the playing field dimensions are known. This is of course assuming that you can accurately detect the angular position of the two beacons separately. I really like the idea of using a stepper motor or even a couple servos to rotate a small phototransistor circuit. As of right now, we have a big arm on the robot so doing a full 360 degrees of rotation would not be feasible with a stepper motor.

I am still left wondering how to differentiate the two signals from each other. Is it as simple as having a phototransistor circuit with a comparator that goes high at a certain threshold. Then I could feed the output of the comparator to an I/O line a microcontroller and check the time that it is high? It just seems a bit too simple to actually work practically.

 

[Nigel Goodwin]

You could have a single sensor, and filter the output through two notch filters, one 2.5KHz and one 4KHz - scan the sensor round in a circle and measure the signal output from each filter as you do - have the sensor in a tube so it gives a fairly narrow angle of detection.

I think I would manually make and test a detection system first, build the beacons as the spec, build a sensor in a tube with filters, and simply move the sensor by hand while monitoring the output from the sensors on a couple of voltmeters (obviously having rectified the outputs first), or use two beams of a double beam scope to monitor the filter outputs directly.

Nice competition rules though, it makes me want to go and build one!

 

[dknguyen]

**broken link removed**

I am still left wondering how to differentiate the two signals from each other. Is it as simple as having a phototransistor circuit with a comparator that goes high at a certain threshold. Then I could feed the output of the comparator to an I/O line a microcontroller and check the time that it is high? It just seems a bit too simple to actually work practically.

THe only problem with that (as I mentioned earlier) is I'm not sure what happens if the sensor sees both beacons at the same time, then you end up seeing some strange superimposing of the two (or beat frequencies or whatever). The beacons are probably not synchronized to each other either. You could code it so that if it was a funny reading that wasn't exactly clear cut, you would assume you were reading both beacons at the same time. Simple is good. Try it out. You might be pleasantly suprised.

 

[Nigel Goodwin]

THe only problem with that (as I mentioned earlier) is I'm not sure what happens if the sensor sees both beacons at the same time, then you end up seeing some strange superimposing of the two (or beat frequencies or whatever). The beacons are probably not synchronized to each other either. You could code it so that if it was a funny reading that wasn't exactly clear cut, you would assume you were reading both beacons at the same time. Simple is good. Try it out. You might be pleasantly suprised.

Hence my suggestion above - you should also bear in mind that the sensors are a considerable distance apart, in the corners of a closed square, and the robot isn't very far away from them. So with the sensor in a tube (giving a narrow beamwidth) it should be impossible to see both beacons at once (remember the walls are black to minimise reflection).

 

[bobledoux]

Set up a sample beacon and try different reception methods. Assuming each beacon is putting out a square wave,

Read the signal from the beacon on a scope. Create a microcontroller program to count the clock cycles that the signal is high. It should be able to set a pin when the signal is high for that frequency. Some sensors may be able to read the signal without need for further amplification.

If precise angle to beacon is required place two sensors, side by side, with a divider between them. Adjust angle and length of divider so both sensors read the beacon only when the divider is pointing directly at the beacon. This eliminates trying to search for the strongest signal strength using a single sensor.

Test your light barrier materials for opacity to IR. Don't assume on this point.