Jump to content
Sign in to follow this  
oxxyfx

First peek at my plane tracker receiver

Recommended Posts

Here is the first peek of my receiver which will track the plane based on the rssi signal strength.

The receiver has 2 video and audio outputs, one for recording, and one for the goggles. The tracking happens in a 170-180 degree rotation driven by a hobby servo.

As you can see the receiver is mounted on a tripod and the servo is mounted such that the antenna will look up about 20-30 degrees.

Ox.

post-6-1182877061_thumb.jpg

Share this post


Link to post
Share on other sites

Still working on the programming. The ground tests are very promissing. Once I get around I will post a little video from behind the reveiver as it tracks the plane.

Share this post


Link to post
Share on other sites

Very nicely done! Always a great feeling to get the basic thing working.

I noticed that there were certain points in the video where it wasn't pointing in the optimal directon (around the 0:48 mark).

You may also want to consider changing the algorithm so it remembers the last change and when it scans, it can scan from a different direction with the best guess. Right now the scan seems to be fixed in a clockwise direction - it makes a quick anticlockwise turn then sweeps clockwise. Maybe you can allow the reverse to happen in some situations.

Do you lose signal when the scanning occurs?

Daniel

Share this post


Link to post
Share on other sites

Hi,

The algorithm is such that it remembers which way it is going, then it jumps ahead in that direction to do the scanning. This maybe was not evident on the video, but it does that. The camera I shot this with is not wide angle enough to let me walk from one side to the other and still be in the picture.

The optimal position if the plane is whithin 60-90 degrees in front of the antenna - and it was still there. The servo only moves the receiver if the signal level changes within a pre-established treshold.

Normally the signal is not lost during the scan, however I just noticed that after I changed the servo to one with higher torque, HS 625MG, now it blacking out the video during scanning. This may be the result of my LM2940 (1A) not being able to supply enough power to the circuit and the servo at the same time. I already ordered some LM1805 - 5V which are 3 amp ldo's and that should be enough power for everything - I hope....

Ox.

Edited by oxxyfx

Share this post


Link to post
Share on other sites

Okay, that's good - it wasn't obvious to me but then I wasn't paying real close attention.

Another thing that might be worth trying is a slow sweep instead of a fast one. If it's slow enough, it can change direction immediately if the RSSI starts going down. Keep up the great work and post some videos of the signal when you get the plane up in the air. Nothing like a real flight to test these things. They have a way of working right on the bench but not in the air. I would be especially concerned about actual RSSI variation throwing the algorithm off - right now it seems the assumption is that RSSI variations are due primarily to positional differences. However in actual flight, there may be other factors, such as orientation, altitude, and so on that may affect signal propagation and thus the RSSI level. I'm just very curious how much those factors will affect the performance of the unit.

Daniel

Share this post


Link to post
Share on other sites

that's interesting but... using a little patch antena with a 60º of beam width is not very usefull,

the goal is to use it with parabolic antenas that just got a beam width of 6 o 7º,

but then a problem appears: image will be null when scanning, and you need pan and tilt,

if you use rssi to point the antena could be possible to add two antenas more at left and right in order just to know if signal is moving to left or right, then move antena in a smoth movement?

well correct me if im wrong, maybe my way is not the best one,

obviously the best way is to use a modem that send gps coordinates to ground and then antena point itself to that coordinates,

Edited by wallaguest1

Share this post


Link to post
Share on other sites

Actually that is a good idea, and one I've been thinking of myself. With two slightly separated antennas, you can make a better guess at the best direction to point the unit, AND you can have diversity reception as well. The only problem is that this will effectively double the cost of the project, result in a bulkier setup, and you still don't have differentiation in the vertical plane. For that, you need a third antenna and receiver!

Daniel

Share this post


Link to post
Share on other sites

Well, the unit is designed to work with a wide beam antenna. I have a flying field where the left - right direction is virtually unlimited, but the depth of the filed is only about 900ft. On this field we mostly fly in rectangular pattern. in order to properly fly fpv, someone would have to orient the antenna even if 20 degrees left or right to cover properly the rectangle. This device is designed to do exactly that.

Imagine that your patch is looking straight ahead and your plane is running out of the beam on the left or right. YOu have to turn the plane around otherwise you'll loose the video.

Using a parabolic or yagi has a narrow beam, but it would not help me in the above situation.

However you are still correct, I am also working on a spatial - 360 degree receiver solution with 4 receivers and 4 patch antennas. That will cover a 4-5 square km area - you being in the middle. This will be a plug and play solution - you could use it with a minimum 2 receivers, or the maximum 4 - autodetected at boot.

Share this post


Link to post
Share on other sites

Here is a second peak to the almost finished receiver. The software is done, tracking nicely - tested on the ground and in the air, 4 different sensitivity settings, 3 delay settings, 3 different angle settings and 2 different calculation for the RSSI signal - average or maximum.

One button operation, 4 standard channels based on the airwave 625 RX.

The aerial can be changed to a more directional one as long it is small and does not weight much more than the GP Patch.

There are some minor video quality issues still left to work out.

To view: Plane Tracking Video 2

or to download: Download

Ox.

Share this post


Link to post
Share on other sites

Nice work there Ox.

I've watched the video and noticed that as you were moving to the right (anti-clockwise w.r.t. tracker) the tracker would make a short clockwise turn before scanning in the anti-clockwise direction.

Perhaps you can consider doing it the opposite way. If I were coding this, what I'd do would be to keep a record of maybe the last 2 position changes and based of that, infer the direction the plane is flying. With that info, I can determine which way to scan which will yield the fastest lock time with minimal dropouts.

So, for example, if the tracker sees the plane going left from the last change, the next scan will be towards the left and scanning clockwise. Ie. a quick anti-clockwise turn (because that is the most likely direction of the plane) and scan back from there.

What seems to be happening in the video is the opposite. As you walked left, the scan seems to start from the opposite side of where you are, possibly causing a dropout before the scanning re-acquires the signal.

I hope I am clear enough - it's a bit hard to explain but since you're so immersed in this you should be able to get the idea. You've got a great platform there - now it's just a question of the best algorithm for best performance. Keep it up!

Daniel

Share this post


Link to post
Share on other sites

I know exactly what you are talking about. I tried both ways, and this way it seems that it does not loose the image it recovers faster - kinda it's going after the plane. The other way would go ahead of the plane.

I do keep track of the previous position - only one and based on that I calculate which way the next move will be. Even if I would keep 2 positions in memory, the chances of getting the direction right is still fifty - fifty. :)

What you are telling me requires only a variablie switch in my program, I can try that again and see if it looks any better.

Thanks for yout input.

Ox.

Share this post


Link to post
Share on other sites

Would love to hear the results Ox.

In the interest of better AI in the tracking algorithm, here's another scheme you may want to just try. The assumptions here are that the plane is unlikely to suddenly go very far from the existing pointer location. So if you can imagine that your plane is at location 5 below:-

0 1 2 3 4 5 6 7 8 9

It is most likely going to be at 4 or 6, thus not requiring a full-scan.

Secondly, with your system, there is a possibility that you will always be caught in the edges so the tracker may be pointing to 3 while the plane is at 5, and when the plane goes to 6, it may point to 4. One way of avoiding this is to actually go ahead of the plane with some predictive input.

So, how about something like this:-

Last best antenna location 3

Plane's last actual location 5

Current best antenna location 4 (as determined from scanning)

Plane's current actual location 6

So, the next predicted best location should be 5.

However, if we go ahead and point it to 6, that'd be even better.

So at the next RSSI drop, instead of initiating a full scan, how about trying a predictive re-positioning at 6 and if the RSSI does not improve, do a reverse scan (5 4 3 2 1 0). If it does improve then just sit there. This way you can have a very fast optimal repositioning and still lose very little by way of scan time.

I'm not sure how responsive the RSSI output is but if your ADC has enough resolution and the RSSI response is not slow (due to smoothing capacitor), there's a chance that this may yield and even better and smarter tracking algorithm

Alternatively you can just try position 6 and if things are not better, try 4. If things still don't get better, perhaps 5 is the answer. One of the problems that will need to be dealt with is the fact that RSSI drop could be due to other factors such as antenna orientation or increasing distance of the plane. Instead of a full scan, you can sometimes just do a small left-right check to see if the RSSI improves. This could allow you to resort to a more comprehensive scan only when absolutely needed, and save battery power in the process.

Daniel

Share this post


Link to post
Share on other sites

Ok, so here is the first video coming back from the plane. At the beginning the tracking is not enabled, but as soon as I pushed the button and the receiver homed onto the signal - the image became chrystal clear.

Please watch this - it wasn't my intention making this video but few minutes after I took off I realized that the receiver is not tracking. I just pushed the button - man - what a difference....

To view: Tracking video from the plane.

To download: Video Download

It is about 80 Meg, for a 5-6 minutes video - but the quality is excellent even at 320x240 resolution.

You cannot see it in the video, but during this flight just before I entered the field the last time - I lost the propeller from the easystar. I guess the motor overheated because I way using 3 cell lipos - and the propeller got de-glued... So I was forced to land the plane - good thing I was just a little above our landing field...

Ox.

Edited by oxxyfx

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×