Camera Traps for Citizen Science

Just to provide a quick update on the project:

• Unfortunately, the project has somewhat stalled since last year, although we are still dreaming of continuing and hacking a stable & cheap version of a camera trap! (Happy to share our code & schemes with everyone interested btw)
• Our last development was that we have switched from using a keychain camera in our Arduino-based trap, as the price/reaction time ratio was really bad. I.e. the really cheap ones (>$5) take ages (~1-2 sec) to wake up upon trigger (so the tiger will be long gone), and we can't keep it on all the time due to the need to save battery time; meanwhile, the expensive ones (>$30) are just too expensive
• We have therefore switched to using an Arducam, which would increase the overall price of the trap ~$10, but might work a lot better
• We have all the parts but we are not currently actively working on it - waiting for a better time to come! (Aren't we all, always?!)

Thanks for everyone still interested in the project - let us know if you'd like to contribute!

As indicated in our first log, we started this project with two potential designs: one based on Arduino (requires more work for the user) and one with Raspberry (less work for the user).

First Raspberry prototype worked very well - however, Raspberry is a greedy little monster that consumes too much battery* !! Jauler has therefore been trying to hack it to make it less energy-greedy, but while this is providing a very resource intensive task (and one of questionable success), and a suggestion during our ThinkCamp Challenge emerged to try a different board, Jauler has now moved on to using this board (STMICROELECTRONICS STM32F429I-DISC1):

He is trying to load Linux on it, and then go from there! TBC!

* one of the goals, that also emerged in collaboration with MammalWeb.org (see our collaboration notes here), is to have the trap active in the field for a month or even longer at a time, while using standard (e.g. AA bateries). That requires a very energy-efficient solution.

MN has been working on the Arduino based prototype, that was initiatily supposed to cost less than $20 and be composed of these components:

MN made an initial working prototype, and we even "trapped" our LOGIN Startup Fair visitors at Technarium booth the other month :p see the cool camera trap pic merges by MN here and here.

The camera we've been using:

By fiddling with some keychain camera variants, however, MN discovered that the cheap versions of the keychain camera are dealing very poorly with sleep-wake cycles, i.e. to save the battery* they need to be programmed to fall asleep, and waking up, in the end, takes at least 4 sec. Needless to say that is waaaay to much for trying to observe wildlife.

MN will share the code & more details about this keychain based prototype soon.

However, we have now dropped the keychain camera idea & moved on to a bit more expensive, yet more standard & accessible solution (especially for people with less electronics & programming expertise), including an Arducam.

We think it will be a lot easier to accomplish quick reaction times with an Arducam, without using up too much battery. Only catch - ~$10 increased camera trap price, although we think it would still be doable under $30.

Our ordered Arducam just arrived, and we will report soon!

* one of the goals, that also emerged in collaboration with MammalWeb.org (see our collaboration notes here), is to have the trap active in the field for a month or even longer at a time, while using standard (e.g. AA bateries). That requires a very energy-efficient solution.

We have realized that multiple groups are currently trying to come up with cheap (and better) alternatives to current camera traps, including various projects in citizen science community, including MammalWeb.org and others (we were contacted by SciStarter too now, who are following our project).

In this regard, and with help of Margaret and Lucy from ECSA (European Citizen Science Association), we decided to run a Camera Trap Challenge at ECSA2016 ThinkCamp in Berlin, 21st May 2016! We ran it together with Pen from MammalWeb.org, and a bunch of local Berlin hackers joined.

Below are the goals we stated for the challenge collectively, and layed out by Pen:

The Challenge

Inspired by great existing projects, the challenge is to develop a fully free (as in freedom) DIY camera trap that meets these challenges:

1. Long battery life (a month or more).

2. Quick trigger speed (preferably <1⁄2 seconds from motion detection to photo acquisition).

3. Night vision (with infrared flash).

4. Lowest possible monetary total cost of ownership.

5. Rugged weatherproof enclosure.

6. Full documentation for all accompanying hardware and software, suitable for at least secondary schools.

7. Practice responsible ethics and citizenship by using and producing only fully free (as in freedom) hardware and software (no binary blobs).

For all of the above, only 7. is a requirement. Participants are of course welcome to tackle just one or all of the items depending on their interests and skill set. In addition, everyone is strongly encouraged to bring their own hardware, tools, and electronics to this ThinkCamp.

Egle & Martynas from our Hackaday Camera Trap project team participated, and Pen led the challenge. Bellow are some of the things that emerged (we did not have much time for actual hand-on hacking, but brainstorming instead):

Notes from Thinkcamp (Pen)

• We looked at naturebytes’ camera, and looks like the biggest issue is the Raspberry Pi, all other components seem good.

• The Rpi is too powerful and drains the battery.

• If you put it to sleep it doesn’t wake up quick enough.

• So maybe we can start by replacing the Raspberry Pi with something (Arduino?) that only does exactly what we want and nothing more.

• One possible end product is something easy to assemble for school kids.

• Another one is just a recipe to print your own PCB and assemble it with other (off the shelf parts).

• We looked at the Moebius camera which is compact and low power, but it doesn’t seem to be very hackable (you can’t flash the firmware) and in its current form it only has USB IO, which has high latency.

• We cracked open a Browning BTC-5 camera trap (it was hard!). Looks like it’s got a SOC that controls everything on a custom board. Items of note:

• There is a light level sensor.

• The PIR sensor doesn’t seem to different from the $2 ones you can get from adafruit.

• The IR flash is just simple array of IR LEDs.

• TODO: Compare the board in the Browning camera to Rpi, figure out what it’s doing.

• TODO: A “table” with two columns: Make We wanna have a recipe and a kit.

• For the recipe, it can more involved hack like 3D print

• For kit, aim for younger kids so it needs to be much easier to assemble to act as a gateway drug into hacking.

• Put down your contact information!

(Egle)

We had 2 of our own prototypes, NatureBytes prototype and a commercial camera. The latter was brutally taken apart [Martynas’ words] by the team (and Martynas later figured out what was broken).

After playing with all the toys, the challenge team also discussed the different features they’d like to have in a camera trap, and how cheap can it be made without compromising the quality of pictures.

In the end the team agreed to continue working on the prototypes, especially the Arduino version, which is the cheaper and more energy efficient alternative. Since we all agree the camera trap should be open source and tinkerable, it will be possible to adapt the Arduino-based core of the trap to be used with different components, such as a better camera. However, that would increase the price of the whole device, so where image quality is less of an issue, cheaper cameras can easily be used.

Both Pen and my team agreed we prioritize the DIY nature & the educational value of the camera trap.

Brainstorming & tinkering at the Camera Trap challenge table

Raspberry prototype on the right & Martynas working on the Arduino one on the left. He managed to burn one keychain camera too! :P

There will be a video soon too - TBA

The main points to take away for our team were: putting more effort into finding a viable option for long lasting batery, while keeping in line with the other goals we stated early on.

We met at Technarium for World Create Day, to hack the first prototypes of the two designs we were considering.

Jauler came up with a nifty code that connects Raspberry Pi to a motion-sensitive house alarm unit and a webcam and captures photos responding to movement, with 1-2 sec delay!

Our code is here: https://github.com/Jauler/CameraTraps

We still need to work out:

1) how to reduce the photo delay

2) a battery and a power saving plan that would be durable enough for hours of observation

by MN is still working on the KeyChain-Arduino version - TBC !

Overall - great progress on World Create Day!!

Came up with two possible designs:

1) Raspberry Pi approach:

• Raspberry Pi Zero - $5
• Simple WebCam - $2-10 (could use old donated ones too)
• PIR sensor - $1.5
• Micro SD card (4gb probably enough) - $2-3 ?
• Battery
• Case
• Optional IR lights

Pros:
+ can connect to any WebCam (or any additional tools if necessary) via USB
+ can be connected straight to a monitor for analysis later
+ would not consume too much battery power
+ can be programmed to stand by (save energy) when there is no action, and woken up again to take the picture
+ easier to reproduce and build yourself

+ can program anything on Linux

Cons:
- difficult to get Raspberry Pi Zero
- would take time to program on Linux

2) KeyChain Camera approach:

• KeyChain Spy Camera - $5
• Arduino that controls the PIR and Camera - $2
• PIR sensor - $1.5
• micro SD card (4gb probably enough) - $2-3
• Extra battery
• Case
• Optional IR lights

Pros:

+ cheap parts

+ easier to get the parts

+ there are also better KeyChains with PIR included etc. – but more expensive

+ easier to program

+ uses less battery life

Cons:

- would need to install a bigger battery

- requires soldering

- photo resolution not so great