Public Lab Research note


Using Mobius camera as a "trap cam" in field deployments

by tonyc | March 22, 2016 18:36 | 2,530 views | 13 comments | #12876 | 2,530 views | 13 comments | #12876 22 Mar 18:36

Read more: unstable.publiclab.org/n/12876


tonyc was awarded the Basic Barnstar by warren for their work in this research note.


What I want to do

Get extended field use from a mobius camera, in intervelometer mode.

My attempt and results

notes on trap cam using mobius

modifying the config file allows you to do intervelometer, but also once your increments are greater than 5 mins, the camera goes into a power saving mode, going to sleep and waking only for the photo.

This allows for extended operation. Using the stock battery, I was able to capture 5 days of images, and the battery still registered as 2 out of 3. The camera ships with a 820mAh 3.7v battery.

There are options to supply more power. One is to connect a larger battery to the board. We currently have some 2200mAh 3.7v batteries, the only problem being they have a full size JST connector, and the Mobius board has a micro JST connector. Trivial to make an adapter, but it is an additional step.

Another option, using stock batteries or otherwise, is to add power by connecting batteries in parallel. I've ordered a cable for this purpose, allowing for 6x JST batteries to be connected in parallel down to one JST plug.

51C7PdehD9L._SL1024_.jpg

The camera can also have power supplied via USB, during normal operation. You can connect an off-the-shelf phone-charger battery pack and provide additional power, beyond the stock battery, via the USB port.

Adafruit sells a power boost board that lets you essentially turn a standard JST Lithium battery into a USB power supply. I've ordered a couple of these: https://www.adafruit.com/products/2030

2030-07.jpg

I'm also wondering if we couldn't provide a trickle charge via Solar, thorugh the USB port. If this worked, the camera could run indefinitely, until the SD card filled up. Right now there is a 32GB max card supported, but with still photos, this is a rediculously long depoloyment.

Questions and next steps

More tests using the ideas above.

Why I'm interested

this would be a univerally cool tool for capturing data. Specifially, thinking this could be used to capture turbidity events in streams, potentially could be deployed in combination with a secchi disk to capture turbidity data, but also contextual data as to stream flow and clarity.

Also, I want to see who is eating my sandwiches in the office.


I did this Help out by offering feedback! Browse other activities for "timelapse"


People who did this (0)

None yet. Be the first to post one!


13 Comments

Cool - it'd be neat to have a script to run on it to generate an image of "change over time" -- i.e. which pixels changed the most. It'd take some doing, but it could highlight things as "tracks" maybe... it'd take some correction for overall light level too.

Reply to this comment...


This is great. Jeff is right that you need a post processing routine to learn if you captured anything interesting. In fact did you look closely at frame 197?

trapcamframe.jpg

I bought a few of those boost converters on ebay where they cost $3.00. So far one of them is junk and one really works supplying 5.2V even from three eneloops (3.6V). The third hasn't arrived yet. I was hoping one would output 6V to run the SkyShield, but the commodity versions are all designed to charge devices with 5V. Pololu has one that will output 6V but it's $12.00. Pololu also has one that outputs 5V for $4.00

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


The settings I used for the above time lapse, from the config file for mobius camera:

Photo Set Time Lapse Shooting=[8],0:off,1:0.25s,2:0.5s,3:1s,4:2s,5:5s,6:10s,7:30s,8:Custom
Photo Time Lapse Shooting Customized Value=[6]    ; in minutes, values from 1 to 43,200 (30 days)
When time lapse shooting set to 5Min or above, Mobius will work in power saving mode (auto turns on/off).

Reply to this comment...


Today I got the solar phone charger and 32GB sd card, so I built a crude outdoor housing and asked for permission to throw this thing on the roof for a month.

I took a clear plastic container and mounted the solar array on the top, with the camera underslung by its mount, using the velcro tape provided with the camera.

IMG_0018.JPG

I used thick baking-rack foil material to make a waterproof stand, which presses the array against the top of the plastic housing.

My thinking is that having it angled ever so slightly downward will keep rain out, and having the camera back a bit will protect against driving rain. The camera isn't waterproof, but its housing is pretty tight and I think this should be enough.

IMG_0022.JPG

In future deployments, a simple front piece of optical acrylic or just picture glass could make it work better, but that would introduce its own issues with fog and moisture clinging to the glass.

It looks as if the trickle charge is happening with no modifications. I did chose the "flip image" setting in the config file, since the camera is mounted upside down.

IMG_0023.JPG

Cross fingers. Do not open until May 4. I'm coming back from a trip that day, and it's star wars day, so let's say may 4th... t

Reply to this comment...


This is so great.

Quick question: I could imagine extending the battery on a Mobius by controlling a USB power source with a switch: i.e., one could supply power only when one wanted to take a photo. If this is a useful idea, then I'm still not sure how we'd turn the Mobius camera on, once it has power. But then, once it's on, I assume that this mode you've dialed in will simply take a photo automatically?

So, I'm imagining a configuration like:

Mobius + switchable power

Protocol:

  1. Turn on the power (with a microcontroller) at a specific time
  2. Turn on the Mobius (somehow?)
  3. Take a photo (because of pre-selected mode, or triggered via USB, or?)
  4. Turn off the power for X minutes [ Repeat ]

Is that doable?

Also: if you have a multimeter, and can sacrifice a 3.7 volt battery, you could hook it up and measure the current (in milliAmps) when the camera is on, and when it's sleeping, and then we could estimate how long field deployments would last in various modes ...

So great that you're digging into this!

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


The new firmware has a hibernate function, so you don't need to do this. You just let the mobius turn itself on and off, and let your additional power trickle charge when needed. That's the idea at least. In application, I'm having issues.

The camera likes to turn on when power is applied via USB. the solar unit I bought is supplying unreliable power, probably because it is cheap crap, honestly.

There is a setting in the firmware that lets you have the camera power on automatically with USB power applied, and also to have it autorecord once power is applied. Both of these are features for people wanting a dash cam in their car (turn on the car, camera is recording, turn it off, it stops).

However, even with these settings set to not let it turn on/off with power, it still seems to do this in photo mode.

I'll play with this some more. I really want to get this done and on the roof because they are about the tear down the building next to us and it would be awesome to capture! t

Reply to this comment...


The new firmware has a hibernate function, so you don't need to do this. You just let the mobius turn itself on and off, and let your additional power trickle charge when needed. That's the idea at least. In application, I'm having issues.

Cool! I mean -- sorry re: the issues :) But this is a neat feature.

There is a setting in the firmware that lets you have the camera power on automatically with USB power applied, and also to have it autorecord once power is applied. Both of these are features for people wanting a dash cam in their car (turn on the car, camera is recording, turn it off, it stops).

Wow -- this is precisely what I'd hope for :) Basically, what I'm imagining is the following scenario:

  • Let's say we wanted to take a photo of a river when we think it's turbid (because another sensor tells us so). I would set up a microcontroller with a turbidity sensor so that when it senses 'high turbidity', it provides power to the Mobius; the Mobius takes a photo; and then the microcontroller turns off power to the Mobius. This will tend both to prolong battery life on the Mobius, and will avoid someone having to sift through 1000's of photos looking for a river that looks turbid :)

However, even with these settings set to not let it turn on/off with power, it still seems to do this in photo mode.

Ah ... and just so I understand: in video mode currently, you can currently enable a feature such that when you provide power to the Mobius (turn it on) it starts recording video automatically? But the still-photo-intervalometer doesn't seem to work the same way?

This makes me wonder (and this is maybe only relevant to my own particular imagined scheme, above): maybe we could control power to the Mobius in 'capture video when powered on' with a microcontroller, and capture short videos when power is provided (in lieu of photos).

For the turbidity application, for example, it might be sufficient to capture only one, clear short video that shows a very turbid river; so, even if the video file size is larger than a photo, we might only need one short video for our purposes ...

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


It seems that the Mobius timelapse features could be all you need to capture turbidity events. Depending on the stream or river, turbidity events caused by rain events generally last for hours if not days. So having the Mobius take a photo every hour could be sufficient to capture these events. That could produce just a dozen photos per day. It would not be necessary to sort them to find turbidity events if the Riffle was recording when they occurred. And sorting through 1000 photos might be one of the easier parts of the whole process of documenting the events.

There might be other types of turbidity events that are short lived, so more frequent photos could be required. But is the general idea that the Riffle results would be the data and the photos would be for PR and supporting evidence? Or will you be leaning on the photos to be data?

Chris

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


I was just loving this even more after thinking more about it. Simply using timelapse is almost the same paradigm as the balloon mapping kit -- solve it with more memory and battery! It's almost like a balloon mapping kit without a balloon :-P -- it's even in a bottle! It's so great in its bypassing of, say, electronics skills or programming.

I had a lot of great thoughts about video analysis -- graphing brightness/color of a selected region over time...

Reply to this comment...


Update on longevity: I inadvertently left a mobius running the intervelometer at every 6mins over the week of staff retreat. I now have 2490 of the most boring photos you've ever seen, but I can tell by the day/night phases that it captured over 10 days of continuous images, and the battery was not done yet. I still have yet to see my patience outlast the battery on a mobius in the deep-sleep mode!

Total use of the SD card was just 1.5 GB, so I think the battery will remain the weak link. This was using the stock battery of 880mAh. Easy to up this with a larger battery, plugged straight into the board.

I believe using solar via some sort of trickle charge, or over USB would be a larger problem to solve. Plugging in through USB has been more problematic in ways I can't even really describe. It seems to prevent camera from running long term by tricking it into rebooting. Will need more investigation to solve this, but the ability to deploy using stock camera is very promising, even right out of the box!

Think I'll let my folks borrow one to do time lapse on some wildflowers as they bloom, just for fun.

Reply to this comment...


Sounds like we'll actually be deploying one of these in the field soon, as part of WI frac sand efforts. Very exciting! I went ahead and bought this $20 silicone housing. First step is to deploy it as a base kit version, stock battery, at 6min intervals (the shortest allowed in power saving mode).

IMG_0044.JPG

It looks like Wonder Woman's jet, doesn't it?

IMG_0045.JPG

Comes with clear acrylic lenses.

IMG_0046.JPG

It's clearly designed for mounting on a helmet, or some less-than-long-term application, as it's open on the bottom. But I'm hoping this will actually survive, maybe with some tape on the bottom to prevent condensation?

The other method is to use clear glass flashlight "lens" as a way to peek out of a waterproof container, such as a mason jar. Mount the clear disk to a hole drilled in the lid, and seal with silicone adhesive.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Here is the rig, using a ribbon cable and a simple trickle solar phone charger to keep it going.

IMG_0048.JPG

Reply to this comment...


Just a catchup here:

I've continued to tinker with the idea, and have a few notes.

The two ways to add power to the camera seem to be through attaching l larger capacity cell directly to the board, via the tiny JST connector, and by supplying power from an external source, via the USB port.

The internal battery provided with the stock Mobius camera is incredibly strong for its size. It allows for roughly ten days of photos, taking it at the highest frequency allowed in the power saving mode. Any setting which takes photos at a longer frequency than every 5 minutes puts the camera into hibernate, so at 6 min. intervals, this gives you 10+ days of operation at room temperature (assuming heat/cold might affect battery life).

A photo every hour would extend the total life of the deployment, but it's not known if it would do so by a factor of 10, or something more or less. This would be a great future test. But of course, testing it directly could take a ton of time. Better to do the test with a meter, and be an engineer about it.

I have so far had issues with trying to extend this via usb power application. I've tried using the unit pictured above, the solar phone charger. It appeared to be working to apply power, essentially recharging the onboard battery. But after leaving the camera unattended for 2 weeks, I found it powered down, and photos had stopped after the same 10 day spell.

Not sure why.

I recently bought a high capacity 10000mha external Li battery and am currently running a long-term test to see if it can outlast the 10 days in this way. My hope is that by using a simpler cell (non-solar, battery only), it may work.

I have yet to try soldering up a cell to connect direct to the board. I am focussing on using non-invasive ways to supplement power to the stock camera, as this would be much more easily replicated, and easier to swap out fresh cells for old ones in the field.

I have also been playing with super-simple waterproof housings for these cameras. I think the winner is the method of taking a known waterproof/watertight container, and applying a flashlight lens (glass circle) using silicone adhesive. The obvious choice is a mason jar or nalgene bottle, but there are many low-cost cases by Pelican, or phone service boxes or other outdoor electrical boxes that could serve this purpose.

I've currently deployed a simple housing made out of a Tyvek postal shipping bag, which I cut out a hole and gooped up a glass lens onto the surface of. Then I rolled the top of the bag like a standard dry-bag and clipped it using a binder. This allowed me to stowe the camera and external battery in a water-resistant way that I hope would allow it to "breath" any condensate or pressure, but still protect the gear. Time will tell, but with this weather, it may be a lot of time.

The advantage of this method is that Tyvek can be drawn on, allowing you to get sophisticated with camo in an urban setting. It doesn't look like an electronics item, just like some vague shape, which would be an advantage in many settings.

I should also mention I'm using a ribbon extension cable to allow me more flexibility in mounting the lens in a housing. This is meant for dashcam use, but is pretty slick for fitting this into tight spaces. It would allow you to, for example, put a lens module in a deal like this:

51jtvRLg__L._SL1000_.jpg

Then have the cable go back to a waterproof junctin box that held the camera/battery. This kind of a thing would disappear in an urban setting, where theft or discovery would be a big threat to getting your data.

I'm fired up about the potential here. I think further tweaks to this basic tool could lead to real victories in the short term. Photographic data is solid, and actionable. Check the latest blog post for more on this idea.

What else do people see here? Anyone have experience with making housings around this camera? If anyone is interested in helping tinker with either the battery/power parts of this, or the housings, please let me know in this thread!

Thanks

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Login to comment.