No longer a hot issue; GPS on open water is dead on accurate...
I need to make a bot seek a beacon. I think I am going to have to concede that GPS will get me within 20' reliably but I can't count on getting closer. For safety's sake, I want to double that but would consider things with a lower limit. Anyway, lets say I get within range and get a "blip" or whatever and can home in following it. What tech will do that?
It is likely to be in rough seas and might be dark.
Both units (the bot and the beacon) are microcontrollers with xBees talking to each other. the xBees are 900 Pro models and RSSI is useless at close range and there are other units on the same frequency nearby. Only the bot can pivot or move purposefully.
If I put one of these (yes, I know it is discontinued but I have some and there is a “same tech” replacement) in a shielded tube, would it only receive when pointed toward the transmitter?
If you could build some sort of directional antenna you might consider mounting three of them in the bow. If all three where mounted parallel (or with a small angle) to each other you could use that to home in on the beacon. If no signal, keep turning the boat until one or more antennas pick up a signal. If the center one has the strongest signal your dead on. If one of the other have a stronger signal adjust the curse.
I take it you mean 3 receivers? Or somehow switch between antennas? You make a good point. I would probably want at least 2 so if I lose the signal I know which way to go to get it again. With one, I would pivot until I found it then lurch in that direction. It also occurs to me that my “tubes” need to look more like flattened funnels so that I won’t lose signals going up and down waves.
BTW, I need to take another idea from you and go with more than one Arduino. I am using all 4 serials and have a couple of I2C devices on the MEGA 2560 driving this thing. That is one reason the Nordic is attractive.
I have long been pondering this exact question. Something more accurate than GPS (or Galileo) but equally robust outdoors. I considered radio signals and directional antennas. But DIY radio technology is quite daunting for my level of expertise. The available options I consider worth mentioning include:
- wifi router with a “cantenna” - Xbee with sufficient shielding to make the signal directional - RF modules like the 433 MHz ones sold by Sparkfun with Yagi
A “cantenna” is often referred to as a wave shaping antenna. Many DIY projects are described online. None of them have beacons as their application.
Until I am confident enough about radio stuff, I would more likely use a visible light omnidirectional beacon with a directional receiver (in the form of a webcam with image interpreting software). AKA “Light House”.
I have a tilt compensated compass. The issue with short distance bearings is motor latency. When I pivot and detect the direction to go in, if I insist on “warp speed now or we’re all dead”, then we’re all dead. I will be a few degrees off before I get moving forward without continuing to pivot and if I bear in the direction to get back on the bearing, by the time I am “on course” I will have moved far enough off course so that the course isn’t correct anymore. It won’t be far off, but what I am working on now is getting better than not far off. Software tweaks to land based test vehicle are increasingly encouraging, BTW. I nay end up happily recanting on not being able to get the last few feet with reasonably priced GPS uints.
An option I am also considering is very short range proximity detection without regard to direction. I can do the “crazy loops” from the GPS data and I am pretty much guaranteed to pass within a short distance of the target and not know it. If I knew it, I could cut power to the engines and probably be close enough. I have see some IR LED beacons that say 7’.
But the point of the post is that I am becoming less sure that I need direction information. My latest attempts to reconcile GPS readings are pretty consistently resulting in close passes. I don’t think I am closing in on a solution to make imperfect GPS locations work by themselves; at the time these close passes occur the data really gives no clues that this was the right one or very nearly so. In fact I am more likely en route between two slightly inaccurate positions. I just need to know I am passing within a short distance so I can stop and/or do a slow small circle like a jet ski that lost its pilot. If I lose proximity, I canl go back to bouncing between waypoints.
How about keeping track of the compass heading on the approach, then when within your tolerance distance just keep the heading, only using the beacon as a ‘you have arrived’
What about an ultrasonic beacon? Have an array of ultrasonic mics on your robot and an ultrasonic tranducer on the beacon.
Ranging is easy, have the becon ping on ultrasonic and rf at the same time, the rf triggers a timer on the bot and measures the time for the sound to arrive.
Direction finding can be tough as you need to process the input from many mics at the same time.
The real question is how much background ultrasonic noise there is? Then its just a matter of getting a more powerfull transducer for the beacon.
You could audiable frequency sound like a fog horn. This might give much better range but may reduce the accuracy.
Another option that might work is sonar. Water/salt proof tranducers and microphones should be off the shelf. Same idea as with the ultrasonics just that sound travels alot faster in water.
OK, I got one of those and the transmitter to go with it. I am receiving at 2400 baud and I send nothing but the character U (0x55, a nice pattern). I queue the last 100 characters received and count the number of Us to estimate RSSI. This works okay for distance, but I haven’t figure out direction yet. I don’t need to triangulate. I am fine with just “in range in front of me” or not because once I have it in range, if it goes out I can pivot until it is back in, go forward as long as it is in front of me, pivot when it isn’t and continue doing this. I will spiral in on the target within a very few turns. My first attempt was to wrap a cup in aluminum foil and put the receiver inside. I figured I would pick up a good signal aimed toward and nothing turned away. I was wrong. It degrades the signal but I get about the same aimed either way. I am just using wire scraps as antennas.
This is very promising. if directional doesn’t pan out, I may be able to do it purely with my dummied up RSSI if I play with antennas and transmit voltage (but I hope not to use too much voltage).
EDIT - Some numbers…
I get 100 lots of the time when they are close together intermingled with high 90s. It drops off pretty quickly into 70s and 60s and then not a lot further it drops into 40s and 50s for a decent range.
