Utilise off the shelf hardware and OSS to create a cheap thermal imaging camera for firefighting use

Similar projects worth following
Thermal imaging cameras are available commercially for firefighting usage. Their ability to see through smoke make them extremely efficient tools for search and rescue operations. However, the commercial models are very expensive (the model currently deployed by the New Zealand Fire Service is around 17k NZD).

For this reason, even well funded services like the NZFS strategically deploy them and not all brigades will receive one. Poorly funded third world brigades/departments will find the pricing entirely out of their league. The idea is not to replace the high end equipment, but rather to make a low end option for usage when the high price tag renders the (clearly superior) equipment unattainable.

The goal is to create a device which costs less than 600NZD (415USD at current exchange rates), and make the plans and software freely available to firefighters worldwide. This is envisioned to be a helmet mounted kit with a heads up display.

Generally, fires don't kill you. By the time it gets that hot, people have usually long since died of oxygen deprivation. The smoke is the killer. During the time lives are still saveable, smoke will often render the area visibility to near zero. Firefighters have several thousand dollars worth of very specialized equipment that allow them to carry out operations in these environments with moderate risk to themselves. But it's a difficult task, and we train to work in blackout conditions, strictly by feel. It takes a pretty fair amount of time to clear a structure like that, and meanwhile any casualties are slowly dying of oxygen deprivation. Speed is of the essence, and it also reduces the danger to the firefighters.

A thermal imaging system allows the firefighter to "see" through the smoke, quickly gain situational awareness, and execute a rescue far faster. It's a great tool. But it's also very expensive. That's what this project envisions rectifying.

The target price point of 600NZD makes it affordable to virtually any fire brigade/department anywhere. And it can be an invaluable tool to enable firefighters to save lives and increase their own safety while doing it. Keeping it an open source project also allows ongoing development. The initial goal is a working prototype allowing simple heads up thermal imaging. But I envision a modular approach, wherein sensors can be added. Heat, gas, what else can a USB dongle sense that may be good to know? Other firefighters will think of it. Mesh networking (like other projects have done). Wifi link outside, allowing the officer-in-charge to observe inside activity. The potential options are endless. And that's the idea. Make a tool by firefighters, for firefighters. And make it cheap and cheerful.

  • 1 × RaspberryPi 1 Model B
  • 1 × Sunglasses personal display
  • 1 × FLIR Lepton module
  • 1 × USB wifi adapter
  • 1 × Pro Audio Microphone Gooseneck

View all 8 components

  • Powering the beast

    ed11/11/2015 at 22:43 0 comments

    Following from the change from sunglasses display to a 2.5" S-Video LCD...

    Power: The LCD requires 7-12vdc. The Pi delivers 5v from pins and that's confirmed by testing to not be enough. With the dev SKROSS Reload 5 I have 5vdc 1.5amp output. This runs the unit nicely (and economically, for long periods of time), but the voltage is too low for the LCD. The sunglasses are powered off USB so no problem there, only the LCD is an issue. Now, the production idea was 6x AA batteries, which output a reasonable 8ish volts. So, the logic was, tap into that directly, and use the cigarette lighter bit as previously tested to power the Pi. Great in theory, but it turns out that the unit I have appears to be a 1amp unit. The addition of a wifi dongle to the Pi pushes consumption over the edge, so it reboots. Presuming a 1.5amp 12v car charger can be found this should be an easy-ish fix. But a bit awkward in the meantime, so will probably continue dev work with the sunglasses display.

  • Workin' for the weekend

    ed11/11/2015 at 22:37 0 comments

    Many changes last weekend. Firstly, got it going with the Ras Pi model 2 B. Dramatically faster than the older Pi 1 B, and also a much smaller form factor due to cabling. End result was a proper full wearable prototype. But resultant changes:

    Gooseneck: I'd previously commented concerns about weight. Confirmed. Too heavy, it's gone.

    Video: Still using the sunglasses personal display for the moment, but going off the idea. They are proving too restrictive. They need to be very close to the eyes, no more than about 1 inch, to work. Further out the geometry stops working. Without the gooseneck flexibility, they're proving quite difficult to work with. They're fine for a proof of concept but I don't think they'll cut it for a production unit. Probably looking at a 2.5in S-Video LCD instead. However, that leads into the next problem, for which I'll make a separate log.

  • Capacitor success

    ed11/01/2015 at 03:21 0 comments

    So, finally put a 22kOhm capacitor in on the CS-CE circuit, to 3.3v pin 17, and it works a charm. Lepton is coming up flawlessly everytime, great result. This fixes a fairly major stumbling block. Still struggling a bit with the Reload5 battery module, but did manage to make it boot off of that today. Need to work on that a bit more-or I guess get some rechargable AAs.

    Next up will be some drilling and bolting in order to get the helmet mount clip on to the assembly. That plus the power resolution should give a prototype good enough to get onto a firefighter to demo. Also on the to-do is put a power switch into the circuit somewhere so the unit can stay sealed up and powered on/off. And then start playing around with software, make a pretty web gui to control things, etc.

  • Spaghetti city no more

    ed09/21/2015 at 20:36 0 comments

    Found a right-angle RCA adapter yesterday, in conjunction with a male-male RCA adapter it removed a heap of cables. Several updated photos are up now, along with a rather poorly drawn artist rendition. Probably pointless given the actual prototype photographs, but just in case the judges are pedantic. :) The battery co advised they're shipping the replacement, so sometime next week I should be able to finally cut the cord. A bit more cable cleanup and some software tweaks and this baby should be ready to wear into some test environments.

  • Stuck on shore power

    ed09/20/2015 at 05:36 0 comments

    Bit of a setback, the nifty rechargeable battery pack is bung. The mfg has agreed to an RMA, but that's gonna take a bit, so appears I'll be tied down for a bit longer. Phooey. Tried to find a short RCA lead, or the bits to DIY one today, but all the local shops are consumer oriented. Will have to hit town tomorrow to source the needed bits from a proper geek shop, then should be able to demo something (albeit with a power lead, bleh). Guess I could just buy some AAs, but as I said that's gonna get real expensive real fast. :) Totally fine for operational use, but 18 bucks worth of AAs just to monkey about and film a video is a bit rich for a poor man like moi.

  • First assembly

    ed09/16/2015 at 10:15 0 comments

    Did a first attempt at assembly today. Generally working but will need to source some shorter cables, it's a bit crowded inside the enclosure. Also not 100% sure about the gooseneck assembly. It's very good, but also a tad heavy, and for a helmet mounted device it may be too much. Nice clean install tho, the display cable routes inside the gooseneck. Will take some pics once I find shorter cables, right now it's spaghetti city in there and looks pretty garbage.

  • Cutting the cords

    ed09/12/2015 at 05:59 0 comments

    Have sourced a wifi dongle and a usb battery pack for dev work. The proof of concept on AAs was good, but it'll get real expensive to buy 6 batteries for every hour of testing. Still a very good solution for production tho. Having the capability just chuck in fresh off the shelf batteries is a lot more foolproof than taking things off the truck and putting 'em on chargers. Wifi + vnc gives the nice-to-have feature of being able to remotely view what the firefighters are seeing. Frankly, in many cases this will be unlikely-as the officer you're far too busy officer-ing to muck about with babysitting the individual crew. But on the occasion you have the time it could be quite useful to transmit the image out, and at bigger jobs where the command unit has arrived it could start to become more plausible (plenty of resource by then, so as an officer you've got more time on your hands).

    Need to tweak energy saver settings so the screen won't blank, but otherwise software is looking pretty good so time to get serious about the physical enclosure. Oh, and sourced the wee helmet mount clip as well. It's designed for flashlights, so will need some hackage, but should result in a very nice mount with the standard NZFS helmet.

  • Battery life proven

    ed08/17/2015 at 04:22 0 comments

    Gave it a run on battery today. 6x AA batteries through a standard 12v car phone charger. Got about 1:15, which is good. Expect the end result may vary some, particularly if/when a wifi interface is included, but this proves the basic necessity is fine. The choice of AA batteries is a pragmatic one; they're available anywhere in the world and easily sourced/replaced. The idea is that this is lowest common denominator; as the project is envisioned as a DIY guide for firefighters (or local helpful geeks), they can always choose to upgrade to a dedicated battery solution and likely greatly improve lifetime (but then require the unit to be plugged in to recharge, which is a problematic choice, things get left behind on station by accident that way). A possible solution there would be a pair of power systems, one charging and one sitting live on the truck.

    With the BA sets we currently carry the maximum time in is about 45 mins, so this will exceed that. Realistically the thermal system is likely most required in the first 20 mins or so, once you get to that point you're quite likely to have initial knockdown and early stages of ventilation, so the thermal for interior searches is plausibly less critical. Still useful for finding hotspots, but that's a less time critical job, so popping out to put in fresh batts would be fine. Most likely the camera would be going back in with a fresh crew.

  • Great scott! It's alive!

    ed08/06/2015 at 07:30 0 comments

    Right, had a couple of probationaries heading to national rookie training (I'm also brigade training officer, sigh) so this languished for a bit while getting them ready for the course. But back into it.

    Curiously enough, despite posts that it works great with the model 2, I've had no joy. Gonna have to figure that out. Meanwhile tho I just went back to the model 1 B, and have a fully working, if cranky, application. I've read that one way around the camera choking on boot is a resistor on pin 26 (cs), may try that route and see if it's reliable. At the moment it happily boots and runs the app, but I need to cycle power to the Lepton to get it up. Have also purchased a couple of 2" tfts, will be experimenting with these vs the sunglass display. Next step is to cram all the loose wires etc into some sort of reasonably usable case, and sort out a battery power supply for the lot.

  • Achievement unlocked: Thermal image on Pi

    ed06/30/2015 at 05:05 0 comments

    Finally had a chance to get the Lepton working with the Pi. Seems to have some dodgy on bootup tho; during the boot process it appears the Pi shoots something at the Lepton causing it to get locked up and require a reset. Obviously no good long term, will likely experiment with the newer model 2 Pi board and see if it also has this issue. Getting away from the RCA jack on the model 1 will be ideal anyway, as the newer 3.5 jack will result in far less cable clutter. May also look at the model2 rev A, as the lower power draw could be advantageous, tho lack of ethernet is slightly questionable from a convenience standpoint.

View all 11 project logs

Enjoy this project?



ed wrote 08/17/2015 at 05:51 point

Not sure where we put licensing info as requested by hackaday rules? Main one is Rasbian, which as a Debian fork will be a combo of BSD and GPL. Also there's the Pure Engineering Lepton code, which is a typical OSS, as long as attribution etc is included. Here's the text:

  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