Thermal imaging cameras are objects of desire for hackers and makers everywhere, but sadly for us they can be rather expensive. When your sensor costs more than a laptop it puts a brake on hacking.
Thankfully help is at hand, in the form of an affordable evaluation board for the Panasonic Grid-EYE thermal imaging camera sensor. This sensor has sparked the interest of the Hackaday community before, featuring in a project that made the 2014 Hackaday Prize semifinals, but has proved extremely difficult to obtain.
All that has now changed though with this board. It features the Grid-EYE sensor itself, an Atmel ATSAM-D21G18A microcontroller, and onboard Bluetooth, but has an interesting feature that, as well as being a standalone device, can be used as an Arduino shield. A full range of APIs are provided, and the code is BSD licensed.
This module is not the highest-spec thermal imaging camera on the market by any means, after all it has a resolution of only 64 pixels in an 8×8 grid. But its affordability and easy availability should trigger a fresh crop of thermal camera projects in our community, and we applaud that.
Thermal camera projects have featured quite a few times here on Hackaday. Some have been based on the FLIR Lepton module, like this one that combines its image with a 640×480 visible camera and another that claims to be one of the smallest thermal cameras, while others have harnessed raw ingenuity to create a thermal camera without a sensor array. This pan-and tilt design for example, or this ingenious use of light painting. Please, keep them coming!
[via oomlout]
They don’t mention the price anywhere in that page
64.95€ at element14.com
64,95€ on Farnell
The distributors say ~$75.
If I recall correctly, FLIR Lepton has a Patent for combining thermal data with standard video.
It may be a bit of a gamble if you sell these or use their IP for promotional purposes, but it is not technically violating IP until the software is loaded.
That patent is nonsense unless they go into specifics about code and hardware choices. I’m so tired of people trying to patent the most basic simplest concepts.
It’s only 22.16USD on the digikey website, although it is “Not recommended for new designs” which usually means it’s mid to end life and stock can get a bit iffy.
I think there may be a replacement part AMG8834 (34 not 32) which is I think a direct drop in, although not stocked yet, same price at digikey.
It is also marked “Part Reqires Manufacturer Approval Prior to Shipping” and I cant find what does that mean
It means that, after you send your order, Digikey (e.g.) has to ask you to fill out a (MSWord) form so they can send it to Panasonic. The form asks what you plan to do with their device and makes you promise not to build weapons and such with it.
I sent mine a week ago and I still haven’t heard anything (yes or no) from Panasonic. I guess they’re not much interested in selling these things.
TI does something similar, but it’s a Web form and takes about 1 minute (I’ve never been “disapproved”), not a week+.
at those prices on might as well buy one of the small FLIR modules, they are several times the price but have an order of magnitude more in performance, plus they will continue to be available.
What is the thermopile array that the HemaImaging guys were using? I think it had a higher resolution but was larger and not surface mount.
Found it. It is a 82×62 array from Heimann Sensor GmbH. Not sure on the cost but probably much higher and more complecated. This panasonic might help to get rid of those PIR sensors that will stop detecting you if you stand still.
“affordable” doesn’t really help me out much
also i would not call 8×8 pix a camera … more of a PIR array …
Panasonic doesn’t call it a camera. They call it a sensor array. I suppose the main advantage it has is the ability to determine motion vectors based on hot spots moving across the grid. So you could tell direction of people entering/leaving a room, etc.
so that’s why they write on their product page: “…Grid-EYE sensor which is a 64 pixel IR camera in a surface mount package measuring 11.6mm x 8mm x 4.3mm, which integrates the MEMS sensor, lens and I2C interface.”
Thumbs up!
There might also be enough resolution to do the reverse: optical mice are pretty low resolution; but high enough that they can accurately detect their own movement if they can safely assume that they are on a static patterned surface.
If you wanted to do a pan/tilt ‘scanning’ type IR camera, that trades speed for resolution and stitches together a high resolution image of a stationary object by moving a low resolution sensor, having 8×8 rather than just a single sensor element might be enough to allow you to use the image data to help keep track of the sensor motion, rather than relying purely on your mechanical system being perfectly accurate.
Certainly looks like a neat toy in any case.
I did some work in the lab comparing the GridEye with FLIR units and it’s a good sensor but calling it a “camera” is a bit of an overstatement (despite what Panasonic’s marketing team has decided) in practical terms since it has such a wide field of view for each “pixel” and so little resolution. If you get a “picture” at all, it will be 64 pixels and the usual game of jogging the sensor slightly then re-reading it to increase the resolution will get tough/impossible because each “pixel” (a thermopile sensor point) will still read the same broad area.
It’s a pretty good sensor though and can be gotten with a nifty evaluation board from Digi-Key (I have no connection with them) that makes it pretty easy to play with. I was using the Digi-Key eval board directly with Matlab rather than the usual interface/PIC board.
There’s a group at UCLA that has done some fun work with it and may save the curious some trial and error (and coding) – a few pictures to illustrate what I mean about the resolution as well:
https://sites.google.com/site/uclaembeddedsystems2014/
Element14 is currently doing a roadtest with this board.
The winners just got announced but there should be some interesting reviews of the board coming up in the not to far future.
https://www.element14.com/community/roadTests/1513
Unfortunately the resolution is so low and the image characeristics so poorly defined that I wouldn’t speak of a camera. It’s a sensor array at best. Seek Thermal is the way to go. If only they would be cheaper.