• Still no ML

    Jacob David C Cunningham • 05/23/2024 at 05:59 • 0 comments

    But the POV footage has been great, used it today, been a while

  • Attempt to reduce power draw

    Jacob David C Cunningham • 04/16/2024 at 03:00 • 0 comments

    Well... idk if this is a pointless thing to do since later I will have to undo it anyway (speed up CPU again).

    I went from 240MHz to 10MHz that did help with power consumption.

    I also added auto-dimming to the LCD after 5 seconds of no touch input.

  • Out again

    Jacob David C Cunningham • 04/10/2024 at 03:24 • 0 comments

    There will definitely be a redesign of the body to address size, staying on the hat better and recently to swap batteries. The LCD sips up power it turns out.

    https://github.com/jdc-cunningham/ml-clip-cam/pull/2

  • Bridged, ran into problems

    Jacob David C Cunningham • 04/04/2024 at 01:32 • 0 comments

    It is using software (ESP32) to hardware serial (RPi).

    Figured out the code to program the touch screen so I made own basic one to start/stop and track elapsed time.

    Soldered it so now it has the top board as initially envisioned.

    It is now much bulkier.

    There is still more work to do but I have all the base parts in eg. bi-directional communication.

    I have some more design/printing to do so it stays on straight on my hat. Next time there is good weather fly and record footage for training the ML model which is an unknown to me at this time.

    Videos below

  • Programmed demo code for ESP32 S3 Dev Board

    Jacob David C Cunningham • 03/28/2024 at 04:22 • 1 comment

    This unfortunately took me way too long.

    I kept running into this problem:

    A fatal error occurred: Failed to write to target RAM (result was 01070000: Operation timed out)
Failed uploading: uploading error: exit status 2

     In the end I got past it by using a Linux laptop eg. Ubuntu on my Carbon X1 gen 6.

    I was trying to program it on my MBP i7 2015 with Monterey... dumped like 3 hrs into this damn.

    But I was able to flash their demo code.

    With this I'll be able to read through their code and make my own UI to talk to the RPi.

    I have to look into pairing the two with serial that would be ideal vs. just button events via GPIO (eg. high/low)

  • At the field

    Jacob David C Cunningham • 03/20/2024 at 07:34 • 0 comments

    Oh man I wish I chose 5 min instead of 15 min increments... (I ran out of time with the AP controller, ran into problems, so made it auto record on boot)

    It was not a complete failure and I had a great flight, also a surprise creature (a hawk moth) graced me with its presence

    Warehouse work has been making me lose weight though that's good

    Sample hand landing video frame

    The hawk moth

    Video, this is processing right now, going to sleep

  • Yes sir

    Jacob David C Cunningham • 03/19/2024 at 17:56 • 0 comments

    Lot of problems, still haven't written the AP code yet, have 2 hours

    First here it is

    Look at this shot mmm

    The chasis breaks here when you're trying to remove the supports

    With a daughter board under the LCD it'll look like this

    Now to program the AP controller then to the field!

  • It's chunky, slow start, distractions

    Jacob David C Cunningham • 03/18/2024 at 20:46 • 0 comments

    It took me a couple of tries to get a good clip design, this was freehanded curves

    Then the clip I had to get right, the connector was in the way so I went with a split design

    I'm printing the body above, it's just over 2 hours

    I'll then assemble this and start programming the AP interface, the display will not be functional yet till I figure out how to program it/use it with the RPi via maybe i2c, if nothing else as buttons

    Measure thrice... print thrice?...

    I really am doing a poor job on this project, I have to get it done and I have other things I'm thinking about (life problems).

    But tomorrow will happen, I will be there at that field.

    Basic power system wired

    It'll look like this plus the bottom clip attached

    I have to reprint it again because I forgot to paste back in the T shape thing that supports the back of the pi... and the pi mounts in the front are too weak/they keep snapping in the process of removing the supporting material... 2.5 hr print gotta do it again...

  • Let's goooo I have seen it, mind's eye

    Jacob David C Cunningham • 03/17/2024 at 22:49 • 0 comments

    When I was screwing around with Brilliant Labs' Monocle I had designed this swivel thing that you could tighten. This clip camera can use the same.

    My eyes are in pain, I probably dragged like 300+ pallets today and covered many miles. Sucks to suck. Phone screen tomorrow though. It's a binge day too went shopping got that good good (food).

    I'm gonna crap out a body tonight since I have to write the code tomorrow for Tuesday

    I can print for another 5-6 hrs. Drinking this right now got it for 50 cents at WalMart

    Alright first I'll prototype the clip, a random S-shaped squighly should work. The form factor is primarily based around the Pi Zero's major dimensions. The Pi Zero W will have a 40-pin female header on it.

    New battery for my caliper too nice.

    I'm actually excited about the potential applications of this thing... you can connect a browser window to the livestream and included it in a video for example or seeing my hands control the sticks as I fly and recording myself taking photos with my other pi cam projects ha.

    Damn I'm struggling, mind hurts

    Ugh... I just feel bad... tomorrow... next update, physical build

  • Provisions

    Jacob David C Cunningham • 03/17/2024 at 03:44 • 0 comments

    Here are the intended parts for the first version

    Mostly power supply stuff (BMS and step up)

    This display is hardcore

    https://www.waveshare.com/wiki/ESP32-S3-Touch-LCD-1.28

    I gotta read up on this, how to program it, I'm expecting it to have serial between the display and the pi... so I can transfer the IMU data while it's recording

    I gotta figure out the graphics library and make a basic interface

    Maybe you don't have to program it that way... you can use the SPI display to output a menu built with python/PIL and then get the touch coordinates to know what was clicked, that would be pretty easy to do, no flashing of the display firmware -- no it does not work that way haha... yeah Arduino or something else

    I was imagining this... like a purpose... to use the IMU values for head tracking and get some kind of insight like climbing/direction idk...

    Still nothing substantial done, today's a work day and I am mentally drained after these 10 hr factory shifts

    In order to ensure that I succeed on Tuesday (this thing is built and it records) I am going to go with the AP approach... the display won't be functional, rather the RPi hosts a web app to control it since that's all familiar tech to me and I can bust it out