[REQUEST] Please add port TensorFlow Lite Micro to the new RP2350 (and old RP2040) microcontroller platforms (e.g. new Raspberry Pi Pico 2 and the original Raspberry Pi Pico + third-party "Picos")?

[Hedda]

Can you please add an official port of the TensorFlow Lite Micro (LiteRT for microcontrollers) library to new RP2350 and old the RP2040 microcontroller platforms from Raspberry Pi?

• https://www.raspberrypi.com/products/rp2350/

• https://www.raspberrypi.com/products/rp2040/

Note that there already looks to be a few unofficial ports of the TensorFlow Lite Micro library for the older RP2040 hardware, however there is no official support upstream for RP2040 or RP2350, (and obviously the newer RP2350 MCU should offcer a lot better performance). I think that RP2350 is also very intesting because it features different CPU-cores that uses ARM and RISC architecture (though developer have to choose which cores to bootstrap).

• https://www.tensorflow.org/lite/microcontrollers

There is LiteRT downstream port for older RP2040 but does't look to be maintained + has not been updated to new RP2350, see:

• https://www.raspberrypi.com/news/a-big-bang-update-for-tensorflow-lite-for-microcontrollers/

and

• https://github.com/raspberrypi/pico-tflmicro

Someone has however submitted a Pico2 patch for RP2350 to that GitHub repository but that pull request has not been merged (again a sign that the downstream repo is not being maintained).

• Updated CMake for latest Pico-SDK, including Pico2 support (issue #18) raspberrypi/pico-tflmicro#19

RP2350 is the successor and next-generation of RPi RP2040, as such the RP2350 is newly released so there is not yet widespread support but RP2040 based development boards are already very popular in schools and STEM education so think they are interesting as alternatives to ESP32 because documentation is often easier as often targeting beginners which in turn means can lead to broader use and use cases.

• https://en.wikipedia.org/wiki/RP2350
  • https://www.raspberrypi.com/products/rp2350/

and/or

• https://en.wikipedia.org/wiki/RP2040
  • https://www.raspberrypi.com/products/rp2040/

FYI, there is not yet an official Raspberry Pi Pico 2 board with Wi-Fi but there are third-party boards with WiFi and LAN, see:

• https://www.raspberrypi.com/news/our-rp2350-partners-made-all-this-excellent-stuff-for-you/
  • https://www.raspberrypi.com/for-industry/powered-by/product-catalogue/?category=RP2350
    • https://ilabs.se/?s=RP2350&post_type=product
    • https://docs.wiznet.io/Product/iEthernet/W6100/w6100-evb-pico2

[Hedda]

By the way, with all the current hype of AI and voice control it would be very cool proof-of-concept and real-world use case if could get Raspberry Pi to then make a demo video showing microWakeWord (micro--Wake-Word, a.k.a. mWW) running natively on RP2350 or RP2040 microcontrollers and maybe post a blog post about it a show case for tflite-micro and Raspberry Pi Pico 2!

• https://github.com/kahrendt/microWakeWord

microWakeWord (also known as "mWW") is a fully open-source wake word detection library made for voice assistant appliance hardware based on MCUs, and it depends on the TensorFlow Lite (LiteRT) library to run fast nativley on low-power microcontrollers:

• https://www.kevinahrendt.com/micro-wake-word

• https://www.youtube.com/watch?v=oSKBWtBJyDE

The microWakeWord was created by Kevin Ahrendt and it enables for example the ESPHome firmware (via microphones) to on-device wake words detections on microcontroller devices. microWakeWord has since been adopted by the non-profit Open Home Foundation who use it on Home Assistant voice control satellite hardware based on Espresiff's ESP32-S3 microcontroller (for its own Assist conversation agent):

• https://www.home-assistant.io/voice_control/about_wake_word/#about-on-device-wake-word-processing-microwakeword

• https://esphome.io/components/micro_wake_word.html

• https://www.home-assistant.io/voice_control/about_wake_word/

• https://www.home-assistant.io/blog/2024/06/26/voice-chapter-7/#big-things-for-microwakeword

• https://www.home-assistant.io/blog/2024/06/26/voice-chapter-7/#microwakeword-and-kevin-join-the-team

If you do please have the Raspberry Pi folks post a new blog post to announce the updated and added RP2350 support!

• https://www.raspberrypi.com/news/a-big-bang-update-for-tensorflow-lite-for-microcontrollers/

I also guess that ESPHome would need to update the RP2040 Platform for it or add another platform?

PS: The microWakeWord project ownership as been transfered to the Open Home Foundation so it will always remain as open-source:

• https://www.cnx-software.com/2024/08/01/the-open-home-foundation-adds-hacs-microwakeword-and-music-assistant-open-source-projects/

• https://www.hackster.io/news/the-open-home-foundation-grows-as-it-names-hacs-music-assistant-and-microwakeword-as-partners-2f60ac645e8f

• https://newsletter.openhomefoundation.org/the-open-home-is-not-for-sale/

[github-actions]

"This issue is being marked as stale due to inactivity. Remove label or comment to prevent closure in 5 days."

[github-actions]

"This issue is being marked as stale due to inactivity. Remove label or comment to prevent closure in 5 days."

[Hedda]

This is still wanted and would think this be of interest since it is a very popular development platform for new developers and beginners starting with microcontrollers as much used in schools or other classes and other education situations, (i.e. RP2040 and RP2350 will in the future likely be one of the more common microcontroller development platforms which many of the next-generation developers got started on).