Setting the Arduino environment for tinyml experiments

Studying recently edx Of HarvardX TinyML 3 - deploying TinyML. This online course greatly reduces TinyML The steepness of the learning curve , It can be used as 《TinyML: Machine Learning with TensorFlow Lite on Arduino and Ultra-Low-Power Microcontrollers》 A useful addition to . The original book is TensorFlow-Lite-Micro Outline of relevant design , But from the time of writing to the time of publication ,TFLM It has been developing , The code in the book will be different in practice ;《TinyML》 Nor did they explain the practical deployment very thoroughly , Personally, I think it lies in TFLM Adopt industrialized pipeline Dependency resolution （ Compilation of different hardware platforms 、 library 、 Tool chain dependency ）;《TinyML》 In the chapter related to hardware platform migration , There is no tool chain from the official industry pipeline Peel it off completely , Also led to the light reading can not be reproduced . Why dare you say that ？ Handlebar TFLM Migration to nrf52840 On the Taobao development board , The book ignores nrf Construction of platform tool chain , And how to deal with tf Design of compiler system linkage . No energy to continue because of work TinyML My hobby .Harvard Online courses give an opportunity to continue , After all, watching video is better than watching code And practice to save a lot of effort .

Get down to business ,《Deploying TinyML》 Use “nano ble 33 sense” Development board and Arduino As the software and hardware environment of the experiment . The individual is VS-Code、Pio The fans of , But stones from other mountains can attack jade ,Arduino IDE It's unique . Recorded in this article TinyML Experimental Arudino IDE Environment settings for .

1. install Nano 33 BLE Sense Of Board File

1 open “Boards manager”.Tools --> board --> Boards Manager

2 Search for " Nano 33 BLE", choice “ Aruduino mbed-enabled Boards" Choose the version ”1.3.1

2. Install the required libraries

Arduino Another advantage of ecosystems is that they provide a large number of libraries for performing various tasks , For example, it interfaces with sensor modules or uses common algorithms to process data .

1 open “Library Manager”.Tools --> Manage Libraries

2 Install the corresponding version of the Library

Library Name	version
Arduino_TensorFlowLite	2.4.0-ALPHA
Harvard_TinyMLx	1.0.0
Arduino_LSM9DS1	1.1.0
ArduinoBLE	1.1.3

3. Set preferences

adopt File --> Preference Set up ：

Show verbose output during: compilation and upload.

Enable code folding.

Display line numbers.

4.Dark The theme

If the Arduino Tired of the same white edit box . You can try Dark The theme . Refer to the following link

https://create.arduino.cc/projecthub/konradhtc/one-dark-arduino-modern-dark-theme-for-arduino-ide-2fca81

The effect is as follows .