PTZ (Pan tlilt zoom) DSLR camera tracker with liftoff detection project

Discussion in 'Rocketry Electronics and Software' started by curtisheisey, May 17, 2019.

Help Support The Rocketry Forum by donating:

  1. Aug 25, 2019 #31

    mikec

    mikec

    mikec

    Well-Known Member

    Joined:
    May 9, 2009
    Messages:
    2,185
    Likes Received:
    171
    Nobody else has ever been able to duplicate that 14 msec number; I have no idea how imaging-resource got it. The actual pre-focused latency is thought to be about 90-100 msec.

    Here's my first sequence of shots at 12 fps using the Sharp GP2Y0D810 sensor and an Aerotech Arreaux on an E15-7. Worked well, but it is a bit of a pain getting the sensor positioned within 4 inches of the rocket, I can see the appeal of the lidar. I may play with the ultrasonic sensor, though I have some doubts about beamwidth, range accuracy, and immunity to the noise of the liftoff.

    arr1.jpg
     
    curtisheisey likes this.
  2. Aug 25, 2019 #32

    mikec

    mikec

    mikec

    Well-Known Member

    Joined:
    May 9, 2009
    Messages:
    2,185
    Likes Received:
    171
    Obviously image analysis can be done at some level on phones, but AFAIK most AR apps are just using the phone's orientation sensors and a simple video overlay. It's not like phone hardware is especially magical relative to what you can do on any other small computer platform, and I find the phone development environments needlessly painful to use (YMMV).

    Tracking a colored blob is pretty easy, but controlling a zoom lens and tracking bright exhaust, then tracking smoke all the way to apogee is many many times harder. But it's a fun problem to think about.
     
  3. Sep 10, 2019 #33

    curtisheisey

    curtisheisey

    curtisheisey

    Forum Supporter TRF Supporter

    Joined:
    Jan 27, 2010
    Messages:
    164
    Likes Received:
    36
  4. Dec 17, 2019 #34

    curtisheisey

    curtisheisey

    curtisheisey

    Forum Supporter TRF Supporter

    Joined:
    Jan 27, 2010
    Messages:
    164
    Likes Received:
    36
    So, I did some more work on the trigger this Fall. A disappointment and a positive surprise.

    The LIDAR is ok for low power detection, but at a high angle in the sky for high-powered rockets, or very bright sky, it gets false positives due to sunlight. So that is a disappointment.

    However, the RF link has worked great and much better than I expected. It is very low latency and allows cameras and triggers to be positioned without stringing wires between them.

    I am going to look at filtering or m of n sliding window detection on the Lidar. Also, I am going to revisit the laser trip detection on the fin. Position it with a goose neck base about the fin and have an RF link to a remote camera(s). The problem with this before was having to string a wire from the sensor to the camera.

    I'm going to write one program that runs both on the sensor module and the camera module. It will have a manual override, so that if the sensor doesn't trigger and the rocket is lifting off, you can trigger it manually. I'm using the Adafruit Feather M0 RFM69HCW Packet Radio - 868 or 915 MHz (3176)

    upload_2019-12-17_8-17-24.png
     
    dhbarr likes this.
  5. Jan 3, 2020 #35

    curtisheisey

    curtisheisey

    curtisheisey

    Forum Supporter TRF Supporter

    Joined:
    Jan 27, 2010
    Messages:
    164
    Likes Received:
    36
    I created a PCB (my first!) for the trigger circuit and RF link.

    Options for different sensor inputs (high-low or low-high trigger).

    Camera and aux output.

    Same circuit and code on another tx and Rx unit.

    Poll feature will query other nodes for connectivity.

    Arm will activate sensor. Subsequent trigger sends trigger message to all nodes.

    Trigger override push button (or foot switch) . Commands all cameras to activate. Nodes do not need to be armed.

    Used kicad for circuit layout. Seeed studios for board


    Now back to the tilt unit, or maybe I will make a Bluetooth link and an app.

    Photo shows armed. Other photo shows trigger and breadboard design node IMG_0778.jpg IMG_0774.jpg
     
  6. Jan 9, 2020 #36

    Chad

    Chad

    Chad

    Well-Known Member

    Joined:
    Jul 23, 2018
    Messages:
    215
    Likes Received:
    72
    Gender:
    Male
    Location:
    Dallas
    man that's cool!

    i wonder if you could use a mic for a trigger, rockets are pretty noisy. Like you would setup the camera pointed at the rocket, detect launch via the sound, then automatically rotate the gimble at a rate matching the expected flight acceleration profile.

    edit: if you got real fancy you may be able to use those directional mics like in an Amazon Echo and use the sound for tracking.
     
    Last edited: Jan 9, 2020
    curtisheisey likes this.
  7. Jan 9, 2020 #37

    curtisheisey

    curtisheisey

    curtisheisey

    Forum Supporter TRF Supporter

    Joined:
    Jan 27, 2010
    Messages:
    164
    Likes Received:
    36
    Definitely worth exploring.

    There are these cheap sensor modules out there that have a comparator and a threshold built it for a digital output. They are like $2-5 on ebay and amazon (cheaper that you can build!!). For example, https://www.waveshare.com/sound-sensor.htm

    There are a whole variety of sensor types available, for example, IR sensor (https://www.waveshare.com/flame-sensor.htm), pictured on the left of my recent post with the photo Google
    IR Infrared Obstacle Avoidance Sensor Module for Arduino


    I have a rev2 pcb board done for the trigger portion. It will be able to accept a variety of sensor inputs, high to low trigger, and has a second input for low to high. The parts will come in and I will start testing. I'm working up a little instruction manual. But I plan to open source it.

    Thanks for your suggestion.
     
    Chad likes this.
  8. Jan 19, 2020 #38

    curtisheisey

    curtisheisey

    curtisheisey

    Forum Supporter TRF Supporter

    Joined:
    Jan 27, 2010
    Messages:
    164
    Likes Received:
    36
    Rev2 board of the camera trigger board/rf link is ready for testing.

    I need to populate an LED D5 (ordered the wrong part), and there is a header for an eventual bluetooth module/ Smartphone app. I need to order the proper on-off switch. Note that the silkscreen for the lipo is backward.

    Thanks to everyone for suggestions. The board can accommodate a variety of liftoff sensors. I started to put together a trade-off matrix. The sensor of choice may depend on which parameters are most important to the user. If one can accommodate a launch rail with a switch built in, or is ok with gluing a foil on your rocket's fins for a pair of alligator or paper clips, those ave very compelling choices. The microphone sensor that Chad suggested is worth a look. You can get one for $5 from China (waveshare, etc), cheaper than you can build.



    Presentation1.jpg thumbnail_1.jpg thumbnail_2.jpg
     
  9. Jan 19, 2020 #39

    curtisheisey

    curtisheisey

    curtisheisey

    Forum Supporter TRF Supporter

    Joined:
    Jan 27, 2010
    Messages:
    164
    Likes Received:
    36
    I plan to opensource the board and software. I'm working on writing up some documentation. I will post details to those interested once more mature, and I have tested out the board more.

    The BOM for a board node using an adafruit RFM95 module 1 mile range, without PCB and bluetooth module is $60. You can Knock $10 off using the RFM65 module (Adafruit 3176) with a 500 meter range and possibly more by not populating the parts of the board you don't care about.

    One board node will give control to a local camera only; Two will give you a sensor and a remote camera over the RF link. Three will add another remote camera or a monitoring node.

    A single board design/software supports separate node functions of sensor, camera trigger and monitor.

    A node has two camera outputs and an aux output (e.g, for the future pan/tilt part, ha).

    There are two sets of sensor inputs, low/high and high/low trigger, each with gnd and +3.3V terminal connections.

    There is an output to turn on a laser or other device.

    The three push button switches are to:
    Arm sensor (once armed a sensor trigger will fire camera)
    Poll other nodes
    Force a camera fire manually without sensor.

    Arm and force trigger have terminal blocks for optional foot switches (or other input control).

    At the moment, unfortunately, the frequency is hardwired in the software. It is crying out for a bluetooth smartphone link for configuration and monitoring. This will be in the future someday
     
    Last edited: Jan 19, 2020
  10. Jan 25, 2020 #40

    curtisheisey

    curtisheisey

    curtisheisey

    Forum Supporter TRF Supporter

    Joined:
    Jan 27, 2010
    Messages:
    164
    Likes Received:
    36
    I got a cheap sound sensor with a LM393 comparator built in from Amazon (5 for $5.49). I must say the sound sensor is looking pretty promising. One can set the threshold so normal talking will not trip it. A clap or low shout will activate. The nice thing about sound is that the sensor orientation toward the rocket is not that critical. Lasers, LIDAR, ultrasound require careful placement and alignment fixtures (aka hassle to set up).

    I will test against real rockets, both Estes and HPR at the next launch. Assuming the latency is low, which it should be if it is only three feet from the rocket, it should be good. Thanks, Chad for the suggestion.
    sound_sensor.jpg
     
  11. Jan 26, 2020 #41

    mikec

    mikec

    mikec

    Well-Known Member

    Joined:
    May 9, 2009
    Messages:
    2,185
    Likes Received:
    171
    Sorry to be negative, but I think pure audio triggering is going to prove complicated, since igniters often pop long before ignition, there can be chuffing, etc. Happy to be proven wrong on this.
     

Share This Page

Group Builder