DroneThermal Analog MAV Cameras, Almost Ready

Hi Everyone,

I'm happy to present the final prototype of DroneThermal micro size analog (PAL/NTSC) thermal camera.
The project started 3 months ago and we have published some samples on

our facebook page on 1 March 2015.

After hundreds hours of code work and design changes, the final prototype is ready for production. We signed the agreements with FLIR for Lepton thermal camera cores as well.

DroneThermal is based on FLIR's low cost 80x60 pixel Lepton camera cores.
Video frame rate and resolution is not much but the price is perfect for civilian small drone hobbyists.
You can pre-order it now, in order to take the advantage of %15 Spring discount.

Composite video signal resolution is higher than Lepton, so we can resize the thermal image and add some extra information into the video output. ( marking hottest and coolest objects etc. )
Unfortunately radiometry(temperature reading) looks nearly impossible because of high math and affects of air temperature, thickness, pressure, reflection rate of the objects and camera body temperature.

I'm working on some cool features , so please share your opinions with me for improvements.

Cheers
Melih

@Nick, I hope I will take a video tonight from my balcony. I'm waiting some parts for real aerial video.

@Roberto, sorry I cannot talk about circuit design details, this is not an open-source project. May be later.

@Sashi & Mark,
I cannot share the original FLIR documentations about telemetry because of NDA.
But, I can say it's nearly impossible with or without shutter. Shutter is not important, you can do that with any flat surface because of very small lens diameter. Flat surface calibration is really simple process and we have a button for modes and calibration on the camera.
The problem is not shutter, the problem is other factors (~10 different factor)
The radiation coming from:

Camera internal structure
Lens
Air
Surface (the object)
Other objects(reflection)

But we want to know only surface temperature.
Air temperature,thickness (distance from the surface) and pressure is important factors. Surface reflection rate is really important. For example, Some materials are transparent, some of them working like a mirror in FIR band.

FLIR sharing large mathematical equations to calculate the right temperature. It's simply impossible without computer processors.(VR's video from raspberry PI) and also FLIR is suggesting 2 different black body(lab equipment) and external temperature sensor for shutter for calibration. After all, all calculations will be rubbish if your propeller changes the camera temperature :)

DroneThermal can show the max-min-spot raw sensor values on the screen. So, you can use these values as reference. But we cannot give you right surface temperature.

FLIR announced some details about radiometric Lepton modules. They will make error correction for camera body and lens radiation but the air and reflections still there.

You need to be a member of diydrones to add comments!

Join diydrones

Comments

Moderator

MarioSpeedwagon August 13, 2015 at 10:46pm

I have a Lepton 25 that just arrived, and the drone thermal board should be at the house tomorrow, both from group gets.
Too bad I'm out of town until Saturday. We're supposed to have a live fire training next week, and I'm trying to get it set up before then.
Marc Dornan March 20, 2015 at 8:49am

Could you fly over the roof of a house at a distance of 5M (15 ft) from the roof? In theory it may be good enough to highlight large damp patches or obvious anomalies requiring closer inspection. Would be very interesting to see that if you could fire up the quad again.
Melih Karakelle March 20, 2015 at 8:13am

@Michael,
Resolution of Lepton is not enough for small animals from a distance. A medium size dog covers only 6-8pixel from 50 meters away.

@Ronald,
Thank you Ronald. I was one of them :)
Distributor

Stefan Olofsson March 18, 2015 at 10:02am

Michael: We have the new version with shutter running on our own solution here. I will try to hunt down some cats and dogs for you :-)
Michael Johnston March 18, 2015 at 6:25am

.....when you have an opportunity .... would be interested to see how a dog, cat or other animals (including birds) appear. Would need the altitude / distance from the animal as well. thank-you.
Melih Karakelle March 17, 2015 at 3:07am

@Roberto
Sorry for my mistake. I was thinking it is a raspberry Pi because of similar interface like Pureengineerin's Lepton code. https://github.com/PureEngineering/LeptonModule

I used Paspberry PI compute module in the first prototypes and results was promising but I didn't chose it because of size and current consumption.

@Lazy, Nice to see you :) Unfortunately day time job,projects and the family taking my all time and my personal projects always missing. :/

@Nick, I uploaded a video from last night, I hope you will like it.
Moderator

Vladimir "Lazy" Khudyakov March 16, 2015 at 2:21pm

Melih, you finish plane? :)))
Moderator

Roberto Navoni March 16, 2015 at 1:11pm

@Marc

yes in our design we evaluate the temperature for each pixel. As told by Nick we are under NDA with Flir ,too about the algorithm for estimate the temperature .

@Nick

we don't use RaspberryPI in our demo but VR uBrain 5.2 it's based on STM32F4 processor with on board nuttx operating systems in our design actually we are using about 50 % of cpu performance , for storing , streaming and evaluate temperature sensor for each point. For this kind of feature we need read data from spi and telemetry from i2c bus.

We are evaluating to implement Video Pal on STM32 for have analog input , too but isn't a priority.

best

Roberto
Melih Karakelle March 16, 2015 at 12:58pm
@Nick, I hope I will take a video tonight from my balcony. I'm waiting some parts for real aerial video.

@Roberto, sorry I cannot talk about circuit design details, this is not an open-source project. May be later.

@Sashi & Mark,
I cannot share the original FLIR documentations about telemetry because of NDA.
But, I can say it's nearly impossible with or without shutter. Shutter is not important, you can do that with any flat surface because of very small lens diameter. Flat surface calibration is really simple process and we have a button for modes and calibration on the camera.
The problem is not shutter, the problem is other factors (~10 different factor)
The radiation coming from:
- Camera internal structure
- Lens
- Air
- Surface (the object)
- Other objects(reflection)
But we want to know only surface temperature.
Air temperature,thickness (distance from the surface) and pressure is important factors. Surface reflection rate is really important. For example, Some materials are transparent, some of them working like a mirror in FIR band.

FLIR sharing large mathematical equations to calculate the right temperature. It's simply impossible without computer processors.(VR's video from raspberry PI) and also FLIR is suggesting 2 different black body(lab equipment) and external temperature sensor for shutter for calibration. After all, all calculations will be rubbish if your propeller changes the camera temperature :)

DroneThermal can show the max-min-spot raw sensor values on the screen. So, you can use these values as reference. But we cannot give you right surface temperature.

FLIR announced some details about radiometric Lepton modules. They will make error correction for camera body and lens radiation but the air and reflections still there.
Marc Dornan March 16, 2015 at 12:08pm

Sashi - looks like you answered my question. I suppose it needs to be shuttered.

of 2

This reply was deleted.