Hello everyone,
I am trying to create a project with IR cameras (I am a beginner-ish)
So when I was researching it seemed to me that IR cameras seem big and bulky and i'm not sure how you would go on to code/ connect them to a circuit etc.
If you need any further details do not hesitate.
What is the application? Is it something a trail camera would accomplish?
I want the IR cameras to capture/ be able to "see" what's in frnot of it.
I suggest to pick an IR camera for which there is a connection tutorial and Arduino example code.
I was thinking of maybe raspberry pi or arduino. (what do you suggest)
Also, many of them are just a tutorial on how to diy an IR camera not how to connect one directly
I managed this with Arduino a laptop and a Microsoft Kinect camera. The open source drivers were difficult to obtain as everyone doing this circa 2012-2014-ish have long since abandoned the projects and a (if i have this right) Apple bought out Primesense, the startup that I think had the open source drivers working in the first place.
If you scour the internet for Arduino and Kinect projects, you may be able to get something working in 2024. I made my app (an interactive game) in 2022 and was limited to running it on a Windows 8 laptop I used solely for the project as the brains for the robot/game.
It was an Arduino/Processing project. The Arduino part was extremely easy compared to getting the camera working and writing the Processing app.
All this to say that yes, it's doable, but if you choose to use Kinect as it has an IR camera and can map a 3D environment, it's pretty onerous and I would be remiss to pretend that a project like this will be easy to the average hobbyist.
What is linked; however, is me testing the initial human skeleton detection as I wrote it into the robot's telemetry for the what would eventually be a Hallowe'en game kids could play outside (robot built on electric wheelchair using two large Nerf blasters if they lost hehehe - the first version of the torso is briefly visible on the floor of the clip and the Kinect is visible before installing it all on my printer in the clip).
https://youtu.be/E9OQu9fl_nQ?feature=shared
I followed the work of Melgar and Diez from the book Arduino and Kinect Projects: Design, Build, Blow Their Minds, 2012.
Note the publication date. This is just how I made a project using an IR camera to a range of about 20 feet. It can be done, but there are probably better, more up to date ways today so if nothing else, hopefully I was able to be encouraging if not helpful in a more practical way.
Good luck.
maybe raspberry pi or arduino
For a camera, choose an MCU with more memory than required to store and/or process images. Therefore, first pick a camera.
Thank you so much.
However, when I first started researching about Kinect it was mostly used for motion detection etc.
I am not sure how I could be able to have it detect anything in front of it at 3 different points (unless I put multiple cameras)
Also just an addition, would Intel realSense D415 camera work?
Neither am I. Maybe a detect with PIR sensors that tells the camera to start recording, like trail cams do (to conserve batteries). My Kinect projects certainly do not detect just anything, not even my cats. The middleware that I had nothing to do with in creating was designed specifically to detect humans (skeleton/joints and simple hand gestures).
Afraid I can't speak to that; nevertheless, whatever you do decide on, the community here will undoubtedly be very interested in seeing your gradual progress (remember, one small step at a time before trying to combine steps).
A few of us have asked what it is you're trying to build. What, specifically is the contraption you're imagining and what is its purpose in the broader sense? Letting us know that will certainly help the helpers here give more specific advice to narrow your search, and in doing so, very likely save you time and money.
In addition, as you progress and hit snags (as we all do - well I do anyway) the community will be able to assist with tips once we can see some concrete progress based on test codes, schematics and the like.
I'll take a look at that camera you mentioned, if I can think of anything useful to you I'll certainly let you know.
Good luck!
For those about to build, I salute you! 
1 Like
Thank you so much!!
I will definetly try to get a hang of IR cameras and intel realsense d415 camera first before trying to jump in all of that stuff.
But I thank you for your assistance so far
1 Like
Hello everyone
So I was checking teh intel camera but it seems extremely expensive. Not sure what you suggest
What is its purpose. I know you want your robot to be able to "see" but since cameras don't actually "see" anything, it will be useful to know the purpose. Obstacle avoidance? Facial recognition? Augmented reality?
Yes, sorry I forgot to mention.
I want it for osbtacle avoidance.
I understand cameras can't "see", I'm just trying to play with them a bit so that they can to an extent
Well, consider that drones don't use their cameras for obstacle avoidance(except some optical flow cameras from what I gather, but I can't speak to their effectiveness or anything else since I don't know).
They use a combination of other sensors to accomplish that, such as lidar, sonar, IMU etc. I don't really know what to suggest in terms of a camera for obstacle avoidance. Even in the best case scenario, consider that you'll likely need more than one. What about reverse or avoiding moving objects from the sides?
I'm not trying to squash your idea, I'm just trying to help you manage your expectations over a range of relevant factors such as price, time, your current skill set and the like.
Trust me when I say I'm well versed in imagining projects (like robots) that turn out to take months or years longer than anticipated (especially when I had no experience), and more than that turn out to cost a LOT more money than I had loosely budgeted.
Whatever type of robot you're planning on building, maybe take a look at what the drone community is doing. The Arducopter guys have been around for years, they have tons of online resources to maybe help you dial in a solution that works best for what you have in mind.
Here's a link that might help you sort of orient yourself in terms of the kinds of considerations you'll need to account for in your build:
Here's a link that discusses some of the challenges, I think, in terms of using an optical flow camera purpose built for obstacle avoidance (note that they themselves are clear that, as of the time of this page, it isn't suitable for ground vehicles)
https://ardupilot.org/copter/docs/common-px4flow-overview.html
Here's a link to how the Arducopter guys implement lidar for obstacle avoidance
https://ardupilot.org/dev/docs/code-overview-object-avoidance.html
And lastly, here's the general page on the Ardupilot/copter/sub/rover/etc that is specifically geared toward using sensor technology in ground based units
https://ardupilot.org/rover/index.html
NOTE: I can't speak to the ArduWhatever project today, I am merely trying to help point you in the direction of resources that should prove useful in your preliminary research. I don't know if the gear or software is still widely available or if the software is obsolete or anything like that; nevertheless, there's tons of info on their pages that should leave you more informed after you check it out.
Good luck.
Why use an IR camera for obstacle avoidance? Visible light cameras work.
The ESP32_CAM costs less than USD $10 and for about $100, you can buy a 3-camera 3D depth mapping module from Luxonis.
1 Like
This works for a mobile robot obstacle avoidance? Is it easy to implement?
1 Like
I found this tutorial useful in getting started with the ESP32-CAM, and RandomNerdTutorials hosts a large number of other interesting and working examples.
I'll leave it to the imagination for how one might use a camera for obstacle avoidance.
2 Likes
This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.