Close
0%
0%

PipeCam: Low-Cost Autonomous Underwater Camera

Low cost autonomous underwater camera for long term deployments and exploration

Similar projects worth following
This projects aims to build low-cost in situ underwater cameras for long(er) term, shallow deployments, from relatively off-the-shelf materials.

The goal of the this project is to prove that "It can't be that hard?"

Project constitution:
- Must be made as cheap as functionality allows
- Must be made of materials and off-shelf parts as far as possible
- Must be easily replicated.

There will be three components to this projects:
- Mechanical hardware
- Electronic components
- Software

Possible applications:
- Biodiversity studies
- Visibility indicator for recreational diving
- Long term time lapses

pipecam_v1.fzz

Functional diagram of the pipecam

x-fritzing-fzz - 7.81 kB - 03/29/2018 at 11:35

Download

takePhoto.py

The main script that handles taking of photos

x-python - 4.11 kB - 03/29/2018 at 11:10

Download

pipecam_v2.fzz

A Fritzing schematic of a Pi Zero + Arduino Nano combo concept for future use.

x-fritzing-fzz - 40.31 kB - 03/29/2018 at 11:36

Download

  • 1 × Raspberry Pi Zero W
  • 1 × Raspberry Pi Camera Module V2
  • 1 × 10mm Plexiglass/Perspex lens
  • 1 × USB Hub pHat
  • 1 × ON/OFF toggle switch

View all 10 components

  • Colour Corrections

    Fred Fourie05/15/2021 at 21:26 0 comments

    Because the current version of the project has no light source, colour correction may be useful in deeper deployments. A simple method for colour corrections was applied that yielded good results. 

    Scripts obtained from http://www.fmwconcepts.com/imagemagick/uwcorrect/index.php was tested, and the following command gave the best results for this case:

    ./uwcorrect -m 2 -b 5 -s 10 target.jpg corrected.jp

    Then this command was used to write a small batch processing script in BASH:

    for f in ./photo/*.jpg
        do f2=${f%.*}_processed.jpg;
        echo "Processing $f as $f2...";
        ./uwcorrect -m 2 -b -10 -s 10 $f $f2
    done

    The results:

  • +3 weeks in the harbour: some results

    Fred Fourie05/15/2021 at 21:08 0 comments

    Recently, I deployed two cameras in the local docks in very shallow (~50cm) water with a 4 hour day/night interval. The camera were not placed and ballasted well so they did move around too much to allow for a decent time-lapse. 

    Here are some of my favourite results from that deployment. 

    The photo above is with the Raspberry Pi HQ camera through a polycarbonate "lens". The discolouration seen at the bottom was an experimental LED. The LED did not yield conclusive results in this test. 

    The photo above is the same scene (roughly) after three weeks in the water. The algae growth on the lens is pretty dramatic.

    The next set of photos are with the raspberry pi v2.1 camera and a different experimental lens and LED:

    These show that the 8MP V2.1 camera still produces very good results (although the polluted harbour is very clear) and puts into question the use of the HQ camera at three times the price. Because of better positioning, these images produced by this camera were far more interesting. 

    Here are some of my favourite animal sightings in the results:

    Yet unidentified juvenile fish 

    Shrimp:

    Swarming Mysid shrimp

    A little crab:

    The results of this deployment has made me confident that all the current hardware performs as it should. The endurance is also very much with the spec of the project.

    I believe that this specific version of the project is about one iteration away from being ready to be Beta tested. 

    More to come in the future.

  • Results: 4 days in the murky harbour

    Fred Fourie03/29/2021 at 17:59 0 comments

    After some more tweaks I recently deployed the camera in the harbour and left it for  4 days. I removed all the images too dark to immediately see and applied some colour correction. (log about colours coming soon).

    Deployment:

    • ~4 days (dark photos removed)
    • Interval: 4 min
    • Location: Harbour, close to locks
    • Depth: ~5m

    Processing:

    • Colour correction: correction script applied
    • Frame rate: 5 FPS 
    • Number of photos: ~344

  • ...More March results

    Fred Fourie03/08/2021 at 19:52 0 comments

    A uneventful time-lapse of some mussels on the pillar of a pier.

    Clearly visible mussel excretion. Night time edited out.

    Interval: 2 min
    Duration: 3 days
    Depth: 30 cm
    Number of frames: 380


  • State of the Project, March 2021

    Fred Fourie03/07/2021 at 13:41 2 comments

    We're back in the water for a new set of tests (finally)!

    In the image above there are a few things to note, one: the nasty reflection caused by the transparent tube and internal mountings, BUT there is also critter, a small swimming crab passing the camera.

    Another shot of a different angle, reflection visible, but a great showcase of the new HQ camera lens set to focus closer.

    Here is a quick update on where I am at:

    1. I'm on the third iteration of a PCB, which was manufactured as V05.03 of the design. Code names Cod (alphabetical fish names for the versions that make it to production). 
    2. PCB design is now two stacking PCBs, one with the MCU and some toys called the "Barnacle board" and the second, a carrier board for the pi, called the "Whale" or "Baleen board".
    3. PCB is now a larger rounder shape.
    4. The PVC tube, although bulky, is still the simplest and most reliable housing
    5. The camera has been upgraded to the Pi HQ camera which has been fantastic
    6. Current version has not been fully optimised as "low power" yet, but is the lowest power version so far.
    7. As I write this the camera is in the water for the second test deployment to try and iron out some more kinks. Yes, there are quite a few. 
    8. User interface has been scrapped. 
    9. LCD screen has been scrapped.
    10. Rotary encoder button has been scrapped.
    11. Project is still made with primary building blocks off the shelf, but the custom PCBs have some extra peripherals.

    There are quite a few design choices to justify, which I will briefly go over here.

    1. PCB size. Below is a photo which compares the sizes in the "Ablacore" version (green, left) versus the latest "Cod" version (black, right). The Albacore was ambitious for my skill level at the time. I spent maybe weeks troubleshooting that board and there where many issues I found to be just too time consuming to fix. I found flashing a bootloader not to be trivial. I took a step back and decided that the space I have in the tube would allow for a bigger board, which makes the design easier to troubleshoot and the PCB shape was made to mound easier in the tube.
    2. Microcontrollers and the carrier board approach. To it's core, this project is a simple intervalometer. There are tonnes of amazing MCUs to use for something like this. I chose to stick with the ATMega328. Because troubleshooting and flashing over bootloaders was too time consuming at the time, I decided to move away from making a custom MCU board, and just going with a carrier board for an off-shelf arduino clone from Robotdyn. Lastly, frying or bricking an MCU does happen and discarding a pcb that was working perfectly otherwise is a pain for this prototyping phase.
    3. Stacking boards. The design changes between versions showed me the need to have some versatility in how the boards can be used and instead of planning for every change. The "Barnalcle" stacking on the "Baleen" has proven useful in the design process. Keeping the two parts of the circuit separate has made troubleshooting relatively easy. Plus it looks cool. 
    4.   Scrapping the UI. The Albacore version had a small screen to give users feedback to the mode and settings of the camera. This meant that a LOT of time was going into designing a control method that was intuitive yet still manageable on the firmware side. Although I enjoyed this, I felt that the amount of time it took to get it "just right" kept the project from getting wet. 
    5. Power pack. Initially I decided that the I would go as simple as possible for the power pack for the project. A few iterations in, I decided that this was not good enough. I've used small lead acid batteries and off-the-shelf powerbanks, but found that the 18650 Li-ion battery tech was very mature and super accessible, which pushed me into building some really basic 18650 rechargeable packs. Also- turns out that I really enjoy power pack design. Just not so good at it.

    I hope to add some more logs and explanations in the coming weeks as I do more tests....

    Read more »

  • Why the project stalled

    Fred Fourie11/26/2019 at 20:29 0 comments

    I was at a very exciting point when I last worked on this project:

    1. I was getting some cool results

    2. I just took delivery of my first PCB for the project.

    3. There was a lot of interest. 

    Then life happened ... HARD. I took a new job opportunity in a different country, uprooted my life, packed, sold my car, gave up rent and had to filter through my worldly possessions for what will make it into my two bags for my massive move.  Many aspects of this project did not make it into the bags. 

    In the time it has taken for me to find my feet the project truly stalled. One big obstacle to getting started again is that I felt that the connection I had with the sea (and my neighbourhood kelp forest) has been lost. And in a big way I am still grieving that loss.

    Without this turning into too big of a therapy session- I am now itching to get back in the saddle, but the pipecam as it is in the current vision may not be enough to explore the sea as it exists in my new home. Finding parts and materials in a new country with a different language has proven hugely difficult.

    Time to rethink and reconsider my approach.

  • An update.

    Fred Fourie09/10/2018 at 05:34 0 comments

    I've been trying to do an honest update on the progress of the project. It's been slow going, but there has been LOADS of progress. 

    At the moment I'm in the trial and error phase of designing a controller board.

    What the controller board does:

    • Control
    • UI
    • Power management

    I'm about a month away from getting back into the water. Of these three aspects, the user interface has taken most of the time. 

    UI

    This has been taking most of my time. If you can avoid it. 

    How users interact with a device is a massive study on its own. I've been juggling giving the users many control options while keeping it as intuitive as possible. This has taken many iterations and hardware changes. It's hard to draw the line between what's possible and what's practical. I think I've finally struck an acceptable balance in practicality/functionality.  At the moment users can set up the device with a dial and a tiny screen for feedback. 

    Control

    The PCB design & etching process is quite rewarding, but it has been a bit of a slow slog. I've made many mistakes in the designs and I'm close to creating a 4th iteration of the board which will hopefully be the last version before I start adding some extras. The design also keeps changing as I learn more about more suitable parts.

    Power

    Power management has been a real eye-opener. There is a lot of control I need to add to reduce the overall power consumption of the project. The power banks I've been so excited about has proven to have some real limitations. This is worth a log in the future, but I've basically decided to scrap them. In short, they are too smart for their own good. 

    I've also come to the conclusion that the project is at a cross-roads. I see this splitting off into two branches. One being the quick and easy one, that folks can build on the cheap and get in the water fast (with some limitations).  Hopefully I'll finish a how-to guide for this in the next few weeks.

  • Feeding Time at the Research Aquarium: Timelapse from Video

    Fred Fourie06/04/2018 at 05:24 0 comments

    I recently got a chance to drop the project in a tank filled with West Coast rock Lobster at the local research aquarium during feeding time. The camera was set to do 15min videos overnight. 

    Here is a timelapse made from the first 15min of footage:

    Problems encountered:

    • Low light (nothing I can do about that at this stage)
    • Camera did not switch over to next flash drive after current one was full
    • Focus was set too far forward to catch all the chaos right in front of it

  • Two drops: Time-lapse vs Video

    Fred Fourie04/25/2018 at 05:50 0 comments

    While I'm still busy with the low power electronics, I'm trying to use the camera to get a feel for the capabilities and limitations... and to figure out how to make it user friendly. 

    So over the last weekend I did two ~5hour drops, both from about 10:00 to 15:00 for the best lighting.

    The first was a 30sec interval time-lapse at the entrance of a small cave. 

    • Depth: ~2.5m
    • Mode: Photo
    • Interval: 30sec
    • Exposure: 'Beach'

    About two hours into the deployment, an octopus discovers the camera, fiddles with it and points it into the cave. Which is both delightful and frustrating. The time-lapse misses most of the action. The pipecamera does not deal with low light conditions that well at this stage :\

    In case you missed it, here is the culprit:


    The rest of the photos taken are very dark, fortunately there are some interesting fish hiding in the cave, but it's too dark to do a time-lapse of these photos. I will consider rather setting the exposure to 'auto' in the future to deal with such... interferances.

    The second deployment was done from 10:30 - 15:30.

    • Depth: ~1.5m
    • Mode: Video
    • Video lenght: 15min
    • Interval: 15min
    • Exposure: Auto

    Because I missed so much action in the time-lapse the day before, I decided to just go with continuous video.

    I placed the camera under a small overhang looking out to a small rocky outcrop. The camera eventually shifts in the surge, which is the reason for the skew video later on. I will need to look into adding a bit more weight in the future.

    At first I thought the footage was pretty boring, but the problem with having 5 hours of footage is that you just skip through it, often missing some of the fun stuff that are only a few seconds long. The little outcrop actually got quite a few visitors.

    After a second round of looking at the results (on recommendation from my video journalist wife), here are some of the highlights:

    A school of fish visiting the deployment:

    A cuttlefish hunting past the frame (LOVE IT):

    Below I took a about a 10min snippet over two videos and converted it into a time-lapse. This is the kind of footage I'm interested in getting more of:

  • Risking the first overnighter

    Fred Fourie04/03/2018 at 05:24 0 comments

    Here is some results and comments on the first overnight deployment.

    Details and findings:

    • 1 minute interval time-lapse was set
    • Pipecam was rested on a small reef on the edge of a kelp forest .
      • ~2m depth
      • Resting with a 3kg weight belt
    • Cronjob set up to operate only between day light hours which is roughly 07:00 - 19:00 at the moment, which looks like this: 
      * 7-19 * * * script.py >> output.txt
    • Was recovered with no damage or leaks
    • The pipecam ran for about 25 hours before running out of power. 
    • The timezone of the pi was not the same as the cron, leading to a 2hour discrepancy
    • About 2.7 Gb of photos was recorded, which in this case was about 660 photos
    • At least two groups of divers found the camera at the location
    • Fish are very interested for about the first hour of deployment

    And here is a video of the results:


    Some conclusions:

    • Needs better identification signage for passing divers
    • Add STATUS Led to easily see that the project is still alive
    • Check timezone and sunrise/sunset times with pi timezone
    • Project is way more power hungry than anticipated (probably the USB hub adding more power than expected)
    • Kelp forests are pretty, but they make for VERY busy time lapses and is better suited for video
    • A night time deep sleep will score another ~12 hours on the current setup

    Now for some bench tests before we put it in the water again. 

View all 18 project logs

  • 1
    Build your own TPL5111 PIPECAM!

    INTRO

    The Pipecamera project has grown into something a bit bigger than what I originally envisioned (Hi-ho scope creep). To accommodate the project growth and the original project constitution I've decided to split the project into two lanes: 

    1. a ready to use Pipecam product with a user interface, field ready and no assembly required, built for long term use.
    2.  a DIY Pipecam, built for rapid, cheap deployment, assembled by the users themselves.

    This post will be discussing the core principals and operations of the DIY TPL5111 Pipecam. I'm not able to do a detailed step-by-step guide at this stage, so this post will aim to give you all the information you require to figure it out yourself. I hope that this will give someone enough information to get them going.  

    I have discussed the housing in the logs before, so this will only deal with the ELECTRONICS and SOFTWARE. 

    ******************************************************************************

    ELECTRONICS

    Parts and what they do: 

    Raspberry Pi Zero W (Any Raspberry Pi will do the job)

    - The main camera controller

    Raspberry Pi Camera Module

    - The actual camera

    Adafruit TPL5111

    https://www.adafruit.com/product/3573

    - The timer circuit

    - Flips a switch ON every X seconds, where X is determined by the resistance set by the onboard pot.

    - X is limited from 1s to 2hours

    - Recommend that you set it to 47kΩ for ~6min (56kΩ for ~10min) interval. See here.

    Adafruit PowerBoost 500

    - The 5V power circuit

    - Handles battery charging

    A battery

    - The power source

    Putting it together:

    ******************************************************************************

    SOFTWARE

    For this, I'm skipping the pi OS install process. I recommend you use RASPBIAN STRETCH LITE and create a default user.

    The pi setup:

    After the OS setup has been completed, install and set up the camera. Find a pretty good guide here:

    https://www.makeuseof.com/tag/set-up-raspberry-pi-camera-module/

    Once that has been done get the operating script here:

    https://github.com/FredFury/pipecam/blob/master/TPL5111_pipecam.py

    Put the script in your directory of choice. Default is: 

    /home/pi/

     Then, add the script to your rc.local. This will run the script on start-up. (ref: https://www.raspberrypi.org/forums/viewtopic.php?t=46113) To do this run the following command:

    nano /etc/rc.local

    Then add the following code before the "exit 0" line:

    string=$(vcgencmd get_camera)
    substring="detected=1"
    if test "${string#*$substring}" != "$string"
    then
        python /home/pi/TPL5111_pipecam.py    # Camera connected start capture
    else
        echo  -e "\e[31mCamera not found!\e[0m"    # Camera not connected.
    fi

    The code above first checks that the raspi camera is present before running the script.

    TPL5111_pipecam.py

    The main operating script will still needs some tweaks and should be considered as under development. The script can be found here:

    https://github.com/FredFury/pipecam/blob/master/TPL5111_pipecam.py

    The script runs a command to take a photo and then shuts down** immediately.

    ** Note that this is NOT a 'safe' shutdown. 

    Python modules used:

    • os
    • gmtime 
    • strftime
    • RPi.GPIO 

    This script should be set to run on boot via the rc.local (or which ever method works best for your application)

    ******************************************************************************


    Workflow:

    1. Once powered ON (battery is connected) the TPL5111 will read the resistance over the pot, which will determine the timer interval.
    2. The TPL5111 waits for the timer to fire
    3. Once the timer fires, the TPL5111 switches the VDD to ENABLE, which causes a 5V output. 
    4. The Pi will receive power and boot up. This takes about 30sec if you've tweaked the boot, otherwise, allow 1-2minutes
    5. A script will run on the pi to take a photo
    6. Once the photo has been taken, the pi flips BCM 17 to HIGH, to let the TPL5111 to go back to sleep. 
    7. Repeat from step 2

View all instructions

Enjoy this project?

Share

Discussions

mchaknov wrote 06/27/2021 at 02:00 point

Hey the github repository doesnt seem to exist anymore, any update on that?

cheers

  Are you sure? yes | no

mchaknov wrote 06/27/2021 at 02:48 point

nevermind i was able to find it

  Are you sure? yes | no

Alex Vila wrote 03/13/2020 at 17:20 point

Do you have links to the screw-on end cap. I'm having trouble locating this part. Thank you. 

  Are you sure? yes | no

Fred Fourie wrote 04/24/2020 at 06:10 point

Here is a link for one 

https://www.pvc24.nl/shop/3-d-koppeling-110mm-2/

Usually they are called 110mm 3 piece PVC union or something along that line

  Are you sure? yes | no

bhaizlett123 wrote 08/05/2019 at 16:24 point

This is one of the coolest projects i;ve seen, i would love to build something like this for deep water, like dropping this down 600 to 900 feet deep, but not sure what kind of enclosure i would need for something like this, as well as what kind of light to put on it, since its pitch black down there, any suggestions?

  Are you sure? yes | no

Fred Fourie wrote 02/17/2020 at 07:20 point

I've not done enough pressure tests with my enclosures. We've dipped it too 100m (~328ft), but it was not a sustained test. Hoping to do some tests with my housing to see what the limits really are. You should probably look into Delrin/POM as a material for an enclosure like that.  You would for sure need a light those depths (It's on my list to build one too).

  Are you sure? yes | no

ahmetyasin98 wrote 06/24/2019 at 16:52 point

Hello, is your camera NOIR, if not, why? Wouldnt it be better for underwater projects where light is low.

  Are you sure? yes | no

Fred Fourie wrote 02/17/2020 at 07:22 point

I have not tested this. IR will get absorbed really quickly in water, so I'm not sure what the use of that would be

  Are you sure? yes | no

Steve R wrote 04/13/2018 at 19:42 point

Awesome project!   Been looking everywhere for something like this.  

Had an idea... have you considered pressurizing the pipe to a few PSI over operating depth?  Maybe something like a needle valve on the bottom that you could attach a tire compressor to.  Not sure if it would have a negative impact on the components.  Benefit would be detecting leaks before deployment, and preventing water ingres if a small leak does occur.  Might make it a little less buoyant, though I think the difference would be miniscule.

  Are you sure? yes | no

Fred Fourie wrote 04/15/2018 at 14:51 point

Hi Steve! Thanks! What would you use something like this for?

At the moment we pressure test each new housing with a modified lens that has a one-way valve that we hook up to a tire compressor. This means we can safely test up to 4bar. The test if considered a success if the housing maintains pressure. 

In a perfect world I would add a BME280 to log the internal pressure over time and temperature, but at I am avoiding the additional expense.  

  Are you sure? yes | no

Giacomo wrote 03/14/2018 at 11:46 point

This is really interesting, especially your progression in battery power and switching to Pi W.  Do you think the Pi W is powerful enough to make the video you need, or will you strip down a Pi 3b+ of the components you don't need (saving power and space)?

  Are you sure? yes | no

Fred Fourie wrote 03/16/2018 at 05:48 point

Pi Zero should be fine to record video. My only concern at this stage is that a Pi Zero + USB Hub = Pi3 with regards to energy consumption. 

  Are you sure? yes | no

danel.turk wrote 03/13/2018 at 08:38 point

Hey

Have you tried to add also motion control to it. To idle on time when nothing happens ?


d

  Are you sure? yes | no

Fred Fourie wrote 03/29/2018 at 12:17 point

Hi there. Motion control is not really suitable for a constantly moving environment such as a shallow seafloor: everything moves.

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates