My main project for 2026 (and onward if it goes well): The PlumeN6 HD: - STM32N6 MCU - MIPI-CSI interface - MJPEG and H.264 encoder - USB-HS - 1000Base-T1 Eth. - ESP32-C5 (maybe) Planned to be used first in either a custom small drone (Andorinha) or a MobED style robot (Lince).

What I'm working on: getting all drivers for the off board sensor set working for the #PlumaN6. This is ICM-45686, BMM350 and BMP581. The ICM over SPI to get the full bandwidth, if required in the future. #drone #diy #flight

Working on the ICM-45686 driver for the #PlumaN6. Who thought it is a good idea to give the register table in big-endian when the DEFAULT is little-endian?? You have to change the endianness to match the register table (set SREG_DATA_ENDIAN_SEL bit).

Interesting finding so far: due to the 32KB Cache, I dont see slowdowns in CPU intense operations (e.g. linear YUV to block YUV for JPEG ENC). Runs at the same speed from SRAM or HyperFlash XIP. More test needed, also for extending Cache and using TCM. #stm32n6 #pluman6 #mcu #arm

It took a LOOONG time but I figured out the issue with my bit errors on the HyperBus over 100 MHz! Besides setting the IO voltage in registers, it is also required setting the OTP fuses! That fixed the issue and now get table 200 MHz both on HyperFlash and HyperRAM! #stm32n6

Been working on the hardware side of the #PlumaN6 and #andorinha drone Project. First version of a 3D printed drone frame, with custom gimbal and articulated arms. Target of around 250g, probably will be over. #drone #3dprintedplane #cad

Designed a new version of my Micro Drill Storage Box. Removed the corner magnets, now with sliding lid, and added drill size labels to each slot! Can be found for free on Maker World: makerworld.com/en/models/280…

I have developed a similar system as well, using 4 ToF sensors from STM to create a 180º Solid State LiDAR. This was part of my Mini Cube Robot project! Had it streaming the point cloud to the PC but stopped there.

I'm starting to look to upgrade my home soldering lab and looking for a Hot Air Station. From what I found so far, the Atten 862D seems to be the pick for great performance for the price. Any other recommendations or experiences with some hot air stations?

Sometimes you have to fit a larger component on a smaller footprint (SOT23 LDO on a WSON-6 footprint in this case). Flipped the LDO upside down and luckily the pin out matches! Faster solution then wait for the correct LDO. #pcb #embedded

Another part of the #PlumaN6 project: I'm looking to design a small DIY gimbal. First version was designed around a cheap 1105 BLDC motor but then I found some better gimbal motors also in 1105 size (smallest I can find)!

First full SD Camera image chain test and working! Fully custom firmware for the #STM32N6 on my #PlumaN6 board. The SD camera chain is: OV7670 -> DCMI -> Line to Block MCU -> JPEG ENC -> USB UVC Buffers are located in the external HyperBUS PSRAM.

Continue work on the JPEG encoder: The standalone (legacy?) JPEG unit is clearly designed for Decoding. The periph. in/out-puts in 8x8 blocks so linear YUV data needs to be transformed into blocks. For DEC, this can be done on the 2D DMA, but on the ENC side only with software..

The final add-on board for the #PlumaN6 (for now). An adapter board for the flexible coax cable to a 24-pin FFC compatible with the Google Coral Camera. This is for the HD camera, using the MIPI-CSI interface.

Navigation for Autonomous Vehicles course! 🚁 @MIT Visual Navigation for Autonomous Vehicles (VNAV) course is available as a free resource covering the entire autonomous navigation pipeline. What's inside? 2D & 3D Computer Vision for navigation, Visual & Visual-Inertial Odometry for state estimation, Place Recognition & SLAM for localization & mapping, Trajectory Optimization for motion planning, Deep Learning for Perception. Complete slides and notes with examples provided. You will learn that visual-inertial odometry combines camera data with IMU sensors for robust state estimation when GPS is unavailable. Obviously you will also learn that SLAM solves building a map while localizing within it. Trajectory optimization ensures planned paths respect vehicle dynamics and obstacle constraints. MIT provides the theoretical foundation for autonomous systems: drones, self-driving cars, mobile robots.... for free. 🎓 ‼️ Save it for later (or start learning today, I recommend this option more): vnav.mit.edu/ ~~ ♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com

Another one of the add-on boards for my #PlumaN6. This is the SD Camera board, designed as a downwards facing sensor set for optical tracking. It is connected to the slow DCMI camera port of the #stm32n6. Besides the camera it also has a ToF and a chip ultrasound (to test).

First picture of the external IMU board for my #PlumaN6 controller board. It has 2 IMU (only one soldered) on individual SPI and power buses, 1 Magnetometer, 1 Barometer (on shared I2C). I also added 4 large resistors with MOSFET control to possibly temperature control the board.

Started to work on Camera/Vision system of the #stm32n6 for my #PlumaN6 board: Discovered that the MCU has 2! JPEG encoders! One standalone, same as in previous STM32s, and one integrated in the VENC. Will start working on the standalone, and on the DCMI camera interface.

First test image over custom USB UVC driver from my #PlumaN6 board. Test pattern generated on the board, not from a image sensor (yet).