Видео

Thanks! This isn't my code, I just followed a tutorial on facial recognition w/ a raspi. I did this a while ago so I forgot which tutorials I used but I think it was here: www.tomshardware.com/how-to/raspberry-pi-facial-recognition or here: core-electronics.com.au/guides/face-identify-raspberry-pi/ or here: ruclips.net/video/P97PRX7ejZQ/видео.html The thing is I'm pretty sure the cmake command from 1 or 2 of these tutorials didn't work and the last tutorial link had the cmake command that worked for me. All of these tutorials pretty much are the same I think. You could find the facial recognition code here as I just cloned it: github.com/carolinedunn/facial_recognition So of course someone named Caroline Dunn made it not me lol For the servo stuff which I added in this video, It's just connected to the gpio pins on the raspi and you can find online how to control servos using the servo library from gpiozero.

And is there a recommended way to send the apriltags position to an esp32

I couldn't really find a complete simple guide for this, so I just googled a lot for library methods. But the way I'm doing it is using OpenCV and the Apriltag library and my motors are connected to a l298n and the enable and input pins directly to my pi 5's Gpio pins. I'm not sure about connecting a Raspi to an esp32 though, I should learn that. I'm guessing you want to use the esp32 because frc will have a teleoperated section not just auto too which your planning to make a custom transmitter and receiver? I'm not sure but I think the pi 5 has built in Wi-Fi and Bluetooth maybe that would work easier? I'm not sure. I just coded this in python using Thonny on the Raspi For detecting the tag: I'm using the 36h11 family which I think is what FRC and all that uses, and I think it's the most popular options = apriltag.DetectorOptions(families='tag36h11') detector = apriltag.Detector(options) so now I can use the detector to start detecting the tags from the options: gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) detections = detector.detect(gray) I'm taking the center of the apriltag by getting the frame width and height and dividing each by 2 for midpoint coordinates on each but flooring it to make it an integer still bc pixels. Just this: frame_center = (frame.shape[1] // 2, frame.shape[0] // 2) where frame.shape[0] is the height and frame.shape[1] is the width as frame.shape gives a tuple for the height, width, and c respectively which I don't know what c is for lol. I'm comparing the frame center to the tag center so if my detected tag's center is to the left of my frame's center then I turn left and same for the right. tag_center = (int(detections[0].center[0]), int(detections[0].center[1])) I get the center of the tag by looking thru the detections list which has all the detected tags (which in my case I'm just looking for 1 so detections[0] only) and I get the x coordinate of the center and the y coordinate of the center too. (.center[0], .center[1]) I created a function to have the logic above if the tagcenter to left of frame center or right and so fourth so then the motors move forward or back for simple differential drive mechanism to turn towards the tag. I'm just calling it then by saying something like followTag(tag_center, frame_center). For drawing the outer box and center I just googled how to you can easily find a tutorial online for that part. I'm not getting all the poses from the tag and I'm only focusing on the yaw related axis of the camera and just relating it to the center of the tag and camera frame.

@@JonathanLi-9813 thanks you this really helps. I should already have open cv installed so I just need to install the apriltag library. Also I was wondering how you would adjust the scale of the tag

@@AussieMich I don't really know how to adjust the scale of the tag in code, as I'm basing it off of pixels instead of real-world measurements. What I'm doing right now is basically saying if the distance from center of tag is >= 50px start turning to realign. If I wanted to make it more sensitive and maybe more accurate I could change 50 px to maybe 40 or something. If I used a bigger tag, it wouldn't really affect it except that it would seem more reactive even with same pixel threshold. Then I could adjust the threshold if I wanted to.

@@JonathanLi-9813 also I’m having trouble installing the apriltag library because they changed how it’s installed and I’m getting a dpkg front end lock error

It's another method of omni directional movement on ground instead of mecanum wheels as people say since mecanum wheel rollers are 45 deg slanted, the force vectors of going sideways n stuff aren't efficient. But that would only need 4 motors, while this is using 2 motors for 1 module, having to turn the module and also having to drive the wheel, so in my case I want 4 modules so I would need 8 motors. I dont really have any data to show that mecanums are less inefficient since swerve uses more motors which means more power consumption but also mecanum needs less motors but less efficient wheels. Crab drive unlike swerve would physically link the 4 turning modules together through something like a chain or what not.

What is the name of this design program?​@@JonathanLi-9813

i

@@JonathanLi-9813 nonono you see i is bad you want 0 i

@@Alex_the_Child713 ok then