Thread Tools
Aug 14, 2018, 04:14 PM
Registered User
barney-NG's Avatar
Thread OP

Update


The version v0.4 will be released in the next days.

What has been changed?

After Simon tested the system with less success I had to upgrade a few things.
  • improve nearby crossing detection.
  • extend resolution to improve far away detection.
  • add some more parameters to change. (Hopefully not needed)
  • now 100 snapshots are stored permanently on the device. (or USB stick if you mount one)
  • 3 beeps once the device is ready for operation
  • added possibility to store motion data. (for debugging)
  • Simon has developed a PCB hat which integrates the piCAMTracker in Simons system! I am sure he will give a report once he's satisfied with the results.
  • Olav made some slightly changes to the cover box

The overall reliability has improved a good step forward. (I am still not at 100% ... but we have some time until 1.0)
I have positive feedback from the F3B community. It seems the v0.4 can be used for speed training.
We made some tests in 20 m/s. If it is not possible to fix the tripod by 100% to the ground detection is not possible. (Below 10 m/s my tripod is sufficient and the detection results are very good)
Last weekend we made tests with a complete timing system. One error invalidates a full run. With this setup the success rate was at ~95%. (I was a bit disappointed because the conditions were not too challenging and error appeared mostly in the same situation and I was not able to find the root cause...)

I think we will never get 100% but I am still confident that it is a good helper for training.

Barney
Sign up now
to remove ads between posts
Aug 14, 2018, 04:26 PM
Team USA F3F Member
ttraver's Avatar
Hi Barney!

I haven't yet been able to put one of your units together and do some testing, but I very much want to.

95% is a good rate, and that's about what we have with the MotCam system. I find that it is approaching the error rate of a human at the base, because there are just as many errors by button pushers.

I hope that I can put something together to help your testing process, as the more testers you have the better the product becomes.

We are having great success running Simons wireless systems here in California, and linking it to f3xvault for pushing scores and all of the data. It's only getting better as we test it, and I'm confident that yours will to!

Have you been able to test it when the sun is directly in the center of the frames? And are there exposure or contrast adjustments that you can make to try and deal with that? That is one thing that we are having troubles with at our local slopes.

Keep up the good work!

Tim
Aug 16, 2018, 09:45 PM
Registered User
SimonWatts's Avatar

piCAMTracker Hat


As Barney mentioned I've worked with him and Olav to develop a plug-in board for piCAMTracker to integrate it to my timing system.

The pic shows the prototype board that is now working successfully. It provides:
- power regulator and shutdown management for the Raspberry Pi
- Xbee on board for direct operation with my timing system
- buzzer for audible indication of turns, power on and shutdown
- expansion header for external turn button and other possible future enhancements
- LEDs for status and xbee connectivity

I have a few changes to do to the PCB design as a result of testing, but it all seems to be ok. I'll work out with Barney and Olav some way of making the final PCB available to anyone who interested - either as a bare board or built up ready to go.

The timing system integration via xbee also offers the possibility of tagging captured turn images with pilot name, time, lap etc, as I already send that data to the base units.

For me the main advantage of piCAMTracker is one small box that has computer, camera and radio network all in one and it talks direct to my timing system.

Now just waiting for some good weather to go an try Barney's detection enhancements

Simon
Aug 21, 2018, 06:06 PM
Registered User

RPLIDAR A2M8 360 Laser Range Scanner


Maybe an option for the near future:

https://www.sparkfun.com/products/14756

RPLIDAR A2 -360° Laser Scanner - The Thinest LIDAR (1 min 22 sec)
Aug 21, 2018, 06:17 PM
Team USA F3F Member
ttraver's Avatar
Unfortunately, probably not.

Laser Radars depend on lasers, "light amplification by stimulated emission of radiation".

Unfortunately you have the biggest laser in the sky to deal with when outdoors, so detection would be extremely difficult...
Aug 21, 2018, 08:26 PM
F3B and F3K
RetoF3X's Avatar

Lidar


Hi Tim

I tend to disagree.

The sun can be considered an incoherent source, whereas a laser is highly coherent and monochromatic. So you can put a lot of energy in a narrow spectral band and detect it with a notch wavelength filter. When you use a laser in the near infrared, it can operate in a spectral range where solar irradiance is pretty low on earth.

I have seen a live demonstration of a LIDAR system, outdoors in bright sunlight. This particular system operates at 1500nm wavelength, has 120x30deg field of view, over 100m range and cm resolution. Such systems are being actively developed for self driving cars.

Not sure what such a unit would cost, but it could distinguish between foreground (grass, bees), the depth in which the F3F plane flies, and the waves further out in the ocean...

Analyzing the point clouds such a system create would be non-trivial, but it is possible to make some sense of a busy street for self driving cars with machine learning and other algorithms in real time.

Reto
Last edited by RetoF3X; Aug 22, 2018 at 09:27 AM. Reason: typos
Aug 21, 2018, 09:39 PM
Team USA F3F Member
ttraver's Avatar
Hey Reto!!!

Prove me wrong!!!!

Please!!! I believe you. Im not actually versed that much in lidar, so I probably spoke too soon.

Guess well just have to mount it sideways and figure out how to discern a plane from a pelican...😜
Aug 22, 2018, 03:46 PM
Registered User
prondel's Avatar
The PiCamTracker in action last sunday at col de mene. The double beep is very helpfull to work anticipation.
At the moment we have a cable and a beep at the center of the course. But we are working on a solution with an android App providing the beep when receiving an UDP frame from the PiCamTracker. This will allow easily to have 2 PiCamTracker sending to the same phone. Once operational, everything will be of course shared on Barney's GitHub.

The PiCamTracker in action (3 min 16 sec)
Aug 23, 2018, 08:21 AM
F3B and F3K
RetoF3X's Avatar
Hi Pierre

Well done. Can you provide details how you have used the TTL signal from the Raspberry board? Did you directly feed it into a speaker wire and then to a buzzer? Or did it need some amplification/processing?

I am contemplating to build a piCAM with a cable too. I do not mind a cable too much, if it has the right length, it can be used to set the distance to the base.

Reto
Last edited by RetoF3X; Aug 23, 2018 at 08:26 AM.
Aug 23, 2018, 02:48 PM
Registered User
prondel's Avatar
Hi Reto !
Hope you are well !

For this session we had 2 buzzers, on at the PiCamTRacker side, one at the center with a 50m cable between.

Below is the cabling of the buzzer.

The second buzzer is connected in // of the first one with the cable between

Cheers,

Pierre
Aug 26, 2018, 03:48 PM
Registered User
barney-NG's Avatar
Thread OP

Progress Report


First of all I like Pierre's video (especially the scene with the camera directly behind the tracker -- I can view that endlessly

V0.4 is out. You can download it here
In the last few days I had so many new ideas that I fear the v0.5 has to come soon.
The most important points:
  • latest Ubuntu release
  • latest firmware upgrades
  • this version is created on a B Plus model
  • debug can be enabled for 30s via web interface
  • debug files are stored together with the stills
  • I found (and removed) the error which broke the object hierachy
  • task already announced: 3 beeps at system startup
  • The "test beep" now creates a picture also. (If you want to place the device.)

Simon, Olav and me we are still working on the PCB to integrate power, shutdown and wireless switch on one easy to solder pcb.
For Simon's problem tracking directly at the slope edge I still have no good answer. (We will try with the camera not turned by 90 degrees next time.)
F3B is working but I need the debugging files of runs where the plane was not detected. (Jeroen's working on that.)
Sren told me that the tracker detected moving clouds on his F3B field. I asked him to increase minimum speed next time he's out. (May be I will use his result as default in v0.5)
In stormy conditions I will try a wind shield in front of the camera the next time.
In Pierre's conditions (bright view, plane distance > 10m, background miles, miles away) the tracker seems to work quite well.

Best regards

Barney
Aug 26, 2018, 03:58 PM
Registered User
barney-NG's Avatar
Thread OP

Sun


Tim -- I didn't forgot your question.

I made some tests with gain and exposure fixing and also with the nd filter.
The good news:
  • The plane is visible in front of the sun!
The bad news:
  • the rest of the picture is really dark.

...we need more time to find a usable compromise...
For this I need to enable endusers to freeze the exposure settings during detection...
(Currently there's automatic gain and exposure because the targets are far away and the camera is fast enough.)

Cheers

Barney
Aug 26, 2018, 04:13 PM
Registered User
barney-NG's Avatar
Thread OP
Quote:
Originally Posted by uzzi
Hello uzzi,

that device is really interesting!
Once these devices are common (payable) we can give it a shot.

Thank you for sharing!

Barney
Aug 27, 2018, 03:12 PM
Registered User
prondel's Avatar
Better than 1 PiCamTracker in action is ... 2 PiCamTrackers in action, with 2 x 50m cables and 2 buzzer at the center of the course:

Practice with 2 PiCamTrackers at each turn (0 min 58 sec)
Aug 27, 2018, 03:17 PM
Registered User
prondel's Avatar
and better than 2 PiCamTrackers in action ... is 2 PiCamTrackers in action connected to an Android mobile app.

OK there is still some work to have an usable App

PiCamTracker used with Android App (1 min 39 sec)


More information here:

https://www.facebook.com/PlanetSoari...4551216282340/
Last edited by prondel; Aug 27, 2018 at 03:33 PM.


Quick Reply
Message:

Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
New Product Valenta Model Mini Twister 3m F3F/B Plane concreteman Slope 3 Jan 16, 2019 12:04 PM
Discussion iMax B6AC detecting 2 cells of 3. Buzzer detects all 3. FromHell Batteries and Chargers 21 Jul 04, 2017 07:56 AM
Discussion F3B/F3F planes timmig Multi Task F3X 0 Jul 07, 2016 10:35 PM
Alert Great Planes Triton Charger does not detect battery! guapoman2000 Batteries and Chargers 38 Feb 07, 2008 09:01 PM
Berg 5 DSP auto detect not detecting rickyblaze Radios 2 Oct 29, 2004 04:29 PM