AI and machine learning
Edge AI and TinyML for Industrial Sensing
Put machine learning on the sensor: tiny models on an ESP32 or Arduino for vibration, current and sound, sent to Node-RED.
Edge AI and TinyML for Industrial Sensing is a free, self-paced online course from EDWartens for instrumentation, maintenance, IIoT and embedded engineers, and final-year electrical, electronics and instrumentation students worldwide who want machine learning on the sensor node. It has 13 modules and takes about 17.8 hours, with written notes, a practice task per module, a practical project and a final assessment of 15 questions (pass mark 60%). Learning costs nothing and needs only a free account. The optional EDWartens Certificate of Completion, verifiable by code, is a one-off US$28.99 for you; it is not a vendor credential. Updated 1 November 2026.
- Free
- Intermediate
- 17.8 hours
- Self-paced
- Certificate optional
- Lab practice in your browser
No card, no trial. The course is free for good. Certificate US$28.99, optional.
- Who it is for
- Instrumentation, maintenance, IIoT and embedded engineers, and final-year electrical, electronics and instrumentation students worldwide who want machine learning on the sensor node
- Jobs
- Roles that use this skill: AI engineer, machine learning engineer, data analyst and automation developer. No course guarantees a job.
- Tools
- Edge Impulse (free Developer plan, in the browser), Arduino IDE 2, Google Colab with TensorFlow and LiteRT, Ultralytics YOLO, Mosquitto, Node-RED and the Wokwi simulator. All have a free route. Hardware: optional. Every module can be done with a smartphone as the sensor, the Wokwi simulator and public datasets. To build the real node: an ESP32 or Arduino Nano 33 BLE Sense with an ADXL345 or MPU6050 accelerometer and a small fan; a Raspberry Pi 5 with an AI HAT or a Jetson Orin Nano is optional for the vision module.
- Updated


After this course you can
By the end you will be able to:
- Decide which plant AI belongs at the edge
- Sample vibration, current and sound at the right rate
- Compute windows, FFTs and spectral features
- Build an anomaly detector on an ESP32
- Quantise a model to fit a memory and latency budget
Who it is for, and what you need first
Who it is for
Instrumentation, maintenance, IIoT and embedded engineers, and final-year electrical, electronics and instrumentation students worldwide who want machine learning on the sensor node.
Level: Intermediate. Intermediate. For instrumentation, maintenance, IIoT and embedded engineers, and final-year electrical, electronics and instrumentation students. You should be able to read a datasheet and edit a short Arduino or Python program. Machine Learning with Python or Predictive Maintenance with Machine Learning is a useful course before this one, but not required.
What you need
Software: Edge Impulse (free Developer plan, in the browser), Arduino IDE 2, Google Colab with TensorFlow and LiteRT, Ultralytics YOLO, Mosquitto, Node-RED and the Wokwi simulator. All have a free route..
Hardware: Optional. Every module can be done with a smartphone as the sensor, the Wokwi simulator and public datasets. To build the real node: an ESP32 or Arduino Nano 33 BLE Sense with an ADXL345 or MPU6050 accelerometer and a small fan; a Raspberry Pi 5 with an AI HAT or a Jetson Orin Nano is optional for the vision module.. What to download, and how.
How this course works
Watch the lessons
Short video lessons, one module at a time, at your own pace.
Read the notes
Open the Reading tab for the written notes and diagrams of the module.
Practise in the 3D labOptional
Try the module in the free 3D lab, in your browser.
Check yourself
Answer the questions to see what stuck before you move on.
Final assessment
Pass the final once you have been through the modules.
Certificate, if you want oneOptional
Optional, with a page anyone can use to verify it.
What you learn
13 modules in order, at your own pace. 17.8 hours of video lessons, with written notes and a practice task in each module.
- 1Edge or cloud for plant AI: latency, bandwidth, reliability and cost32m
- 2The hardware map: microcontrollers, Raspberry Pi with an AI HAT, and Jetson34m
- 3Collecting sensor data: accelerometer, current clamp and microphone50m
- 4Features for signals: windows, FFT and spectral features46m
- 5Training a tiny neural network for the LiteRT microcontroller runtime, with lab practice: 47m
- 6Edge Impulse end to end: data, impulse, test and deploy52m
- 7Anomaly detection on a motor or fan with an ESP32, with lab practice: 1h 12m
- 8Sound classification for machine states42m
Show the other 5 modules
- 9Quantisation, memory and latency budgets1h 36m
- 10Vision at the edge: YOLO on the Raspberry Pi AI HAT, Jetson and ESP32-CAM47m
- 11Getting results out: MQTT, Modbus and Node-RED dashboards30m
- 12Field deployment: power, enclosures, updates and drift38m
- 13Final assessment30m
Software you need
What to download, where from, what it costs and how to install it. Every link goes to the maker's own site, never a mirror.
No hardware is needed: Edge Impulse takes data from a phone, and the Wokwi simulator at wokwi.com runs an ESP32 with MQTT in the browser. Add the ESP32 boards package in the Arduino IDE Boards Manager to build a real node. Ultralytics YOLO is only for the vision module.
Show what to download (7 tools)
Required
- 01Free Developer plan, 3 private projects
Edge Impulse Studio
Edge Impulse (a Qualcomm company), in the browser
- Runs on
- Any modern web browser; the optional command-line tools need Node.js on Windows, macOS or Linux
- Account
- A free Edge Impulse account
The Developer plan is free for individual developers, students and universities: 3 private projects and unlimited public ones, 60 minutes of compute per job, GPU training and up to 3 collaborators per project. It covers learning, internal R&D and pre-production; production use needs the paid Enterprise plan. Terms change, so read the current pricing page.
Steps
- 1.Open edgeimpulse.com and sign up for a free account.
- 2.Create a new project and choose the kind of data (accelerometer, audio or images).
- 3.Collect data from your phone by scanning the QR code under Data acquisition, or upload files, or connect a board with the data forwarder.
- 4.Design the impulse, train, test, then open Deployment and export an Arduino or C++ library.
- Do not upload data your employer has not cleared for an external service; public datasets work for every module.
- The command-line tools (npm install -g edge-impulse-cli) are needed only for the data forwarder from a board.
Open Edge Impulse Studioedgeimpulse.com - 02Free
Arduino IDE 2
Arduino
- Runs on
- Windows 10 or later, macOS, Linux (AppImage and ZIP)
- Account
- None needed
Free, open source; the source code is published on GitHub.
Steps
- 1.Open arduino.cc/en/software and download Arduino IDE 2 for your operating system.
- 2.Install it and start it.
- 3.For an ESP32, open File, Preferences, and add the Espressif boards URL given in Espressif's arduino-esp32 documentation; then install esp32 in Boards Manager.
- 4.Connect the board by USB, choose it under Tools, Board and Port, and upload the Blink example to check the set-up.
- 5.To run an Edge Impulse model, use Sketch, Include Library, Add .ZIP Library with the library exported from Edge Impulse.
- If the port does not appear on Windows, install the USB to serial driver for your board's chip (CP210x or CH340).
- The legacy IDE 1.8.19 is still offered for old computers, but the course uses IDE 2.
Official download pagearduino.cc
Optional
Useful, not needed to finish the course.
- 03Free, GPU time not guaranteed
Google Colab
Google, in the browser
- Runs on
- Any modern web browser
- Account
- A free Google account
Colab is free to use. In the free version GPUs and TPUs are heavily restricted and not guaranteed, sessions can run for at most 12 hours, and idle sessions are stopped. Paid plans give more reliable access.
Steps
- 1.Open colab.research.google.com and sign in with your Google account.
- 2.Click New notebook (or open the notebook your course links to).
- 3.For a GPU, choose Runtime, then Change runtime type, then pick a GPU if one is offered.
- 4.Type code in a cell and press Shift and Enter to run it.
- Save a copy to your Google Drive so your changes are kept.
- Files on the runtime are deleted when the session ends. Save outputs to Drive.
Open Google Colabcolab.research.google.com - 04Free
TensorFlow
Google (TensorFlow project), a Python or npm package
- Runs on
- Python 3.10 to 3.13. Ubuntu 16.04 or later; macOS 12 or later (CPU only); Windows 7 or later (CPU only); Windows 10 build 19044 or later through WSL2 for GPU.
- Account
- None needed
Free, open source under the Apache 2.0 licence.
Steps
- 1.Install a supported Python (3.10 to 3.13) and create a virtual environment.
- 2.Upgrade pip: pip install --upgrade pip
- 3.For CPU: pip install tensorflow
- 4.For an NVIDIA GPU on Linux or WSL2: pip install tensorflow[and-cuda]
- 5.Test it: python -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"
- TensorFlow 2.10 was the last release with GPU support on native Windows. For a GPU on Windows, install inside WSL2.
- If you have no GPU, use Google Colab to train larger models.
Official download pagetensorflow.org - 05Free
Eclipse Mosquitto
Eclipse Foundation
- Runs on
- Windows (64-bit and 32-bit installers), macOS (Homebrew), Linux (Ubuntu PPA, Debian, Raspberry Pi, Snap)
- Account
- None needed
Free, open source under the Eclipse Public License and Eclipse Distribution License.
Steps
- 1.Open mosquitto.org/download.
- 2.On Windows, download the 64-bit installer and run it. On macOS run: brew install mosquitto. On Ubuntu run: sudo apt-add-repository ppa:mosquitto-dev/mosquitto-ppa, then sudo apt-get update and sudo apt-get install mosquitto mosquitto-clients.
- 3.Start the broker by running: mosquitto -v
- 4.In a second terminal, subscribe with: mosquitto_sub -t test, and in a third publish with: mosquitto_pub -t test -m hello
- By default Mosquitto 2 only accepts connections from the same computer. To allow other devices, add a listener and authentication settings to mosquitto.conf.
- Use a lab network only. Do not expose an open broker to the internet.
Official download pagemosquitto.org - 06Free
Node-RED
OpenJS Foundation (Node-RED project)
- Runs on
- Windows, macOS and Linux (needs Node.js 22 or later; Node.js 24 recommended), or Docker
- Account
- None needed
Free, open source under the Apache 2.0 licence.
Steps
- 1.Install Node.js LTS from nodejs.org.
- 2.On Windows run: npm install -g node-red (on macOS or Linux run: sudo npm install -g node-red).
- 3.Start it by running: node-red
- 4.Open http://localhost:1880 in your browser to use the editor.
- With Docker, run: docker run -it -p 1880:1880 --name mynodered nodered/node-red
- Leave the terminal window open while you use Node-RED.
Official download pagenodered.org - 07Free
Ultralytics YOLO
Ultralytics, a Python or npm package
- Runs on
- Windows, macOS and Linux with Python 3.8 or later and PyTorch 1.8 or later
- Account
- None needed
Free under AGPL-3.0 for learning, research and open source projects, but any project using it must also be open source under AGPL-3.0. Commercial or closed-source use, including internal company tools, needs a paid Ultralytics Enterprise licence.
Steps
- 1.Install Python and, if you have an NVIDIA GPU, install PyTorch first from pytorch.org for your CUDA version.
- 2.Run: pip install -U ultralytics
- 3.Test it: yolo predict model=yolo26n.pt
- 4.Find the results in runs/detect/predict.
- The first run downloads the model weights, so you need internet access.
- If you plan to use YOLO at work, read the licence page before you start.
Official download pagedocs.ultralytics.com
Checked against each maker's own page on 27 September 2026. Trial lengths and editions change; the maker's page is the final word.
Practise in the 3D lab
Practise the checks a small model needs before it goes on a sensor: spotting overfitting, reading precision and recall, and setting a threshold that keeps false alarms low.
- Overfitting: tune the treeAI Workbench, about 20 minOpen in the lab (opens in a new tab): Overfitting: tune the tree
- Choose the metric and the thresholdAI Workbench, about 25 minOpen in the lab (opens in a new tab): Choose the metric and the threshold
Optional. Skip it whenever you like: it never blocks your progress or your certificate.
Course project, about 18 hours
An ESP32 vibration node that learns normal on a fan, flags anomalies and reports to MQTT and a Node-RED dashboard
Build a vibration node that learns what a healthy fan looks like and flags anything different. Test it with induced faults and a 24-hour false alarm run, publish its score over MQTT to a Node-RED dashboard, and write the deployment sheet. A phone, the Wokwi simulator and a public dataset replace the hardware if you have none.
- 5documents in the pack
- 7steps to follow
- Laboptional: in the 3D lab
- 7things to hand in
- 18 hof work, at your pace
- URSRequirements for the fan vibration anomaly node
- SDSDesign note for the fan F-07 anomaly node
- Test reportAcceptance test of the fan F-07 anomaly node
- ProcedureCommissioning, update and monthly drift check for the fan anomaly nodes
- Risk registerRisk register for the fan anomaly nodes
Do it in the lab: 4 steps.
The project is optional practice. The certificate needs only the modules and the final assessment.
After the course, your next step
Show what you can do, then keep going in the same field.
A certificate you can verify
Finish the modules and pass the final assessment (15 questions, pass mark 60%) and you can add a certificate with a public verification page. One-off, US$28.99 for you. It is not a vendor credential.
Want a free certificate? Do the matching Free internship: 4 weeks, nothing to buy, with a free certificate of completion.

PLC Programming
Eight short weeks of PLC ladder logic on free simulators: notes, diagrams, a few videos and auto-tested tasks. Finish weeks 1 to 4 and the final, then claim a free Internship Completion Certificate with your real dates and study hours, and a QR code anyone can scan to verify it. Training, not employment.
See the internshipMore free courses
Roles that use this skill
- AI engineer
- Machine learning engineer
- Data analyst
- Automation developer
These are the kinds of roles where the skill is used. We make no promise of a job or an interview.
Plain notes
- Free: the whole course, the notes, the practice tasks, the final assessment and support from practising engineers.
- Paid, and optional: the course certificate, and the Engineer-verified internship (the Free internship costs nothing).
- No job promise: the course teaches a skill. It does not place you in a job.
- Lessons by: Edge Impulse, DigiKey, IBM Technology, EDGE AI FOUNDATION, Andreas Spiess, Hardware.ai, Core Electronics, Edje Electronics, Jeff Geerling, Joyce Lin, Efficient NLP, Circuit Digest, ThinkIOT, Learn Embedded Systems, educ8s.tv, Evidently AI. Independent creators; the lessons stream from their own channels. EDWartens wrote the study plan, notes, practice tasks and assessments.
Edge Impulse · DigiKey · IBM Technology · EDGE AI FOUNDATION · Andreas Spiess · Hardware.ai · Core Electronics · Edje Electronics · Jeff Geerling · Joyce Lin · Efficient NLP · Circuit Digest · ThinkIOT · Learn Embedded Systems · educ8s.tv · Evidently AI. If you are one of these creators and want a lesson removed or credited differently, write to info@edwartens.com.
Practise what you learn in the free 3D lab
Technical work is judged on what you can do: wiring, commissioning, finding a fault. Doing a task yourself teaches more than watching it a second time, and a lab lets you try it safely before you meet real equipment. It does not replace working on real hardware.
- Practise safely, so a mistake costs nothing.
- The lab checks your work like a site test, so you know when it is right.
- Free, in your browser, with nothing to install.
Why learn with EDWartens
What you get with this course, and what comes after it.
- Free from the first lesson to the final assessment.
- Practice in a free 3D lab in your browser, with nothing to install.
- Video lessons with written notes and a project.
- An optional certificate that anyone can verify online.
- Free internships to keep practising afterwards.
Common questions
Who is this edge AI and TinyML course for?
It is for instrumentation, maintenance, IIoT and embedded engineers who want machine learning on the sensor itself. It is also for final-year electrical, electronics and instrumentation students anywhere. It suits anyone who looks after motors, fans and pumps. They get early warning of faults without streaming raw data to the cloud.
Do I need to buy hardware?
No. Every module can be done with a smartphone as the sensor in Edge Impulse. You also use the Wokwi ESP32 simulator in the browser and public datasets such as the CWRU bearing data and the MIMII machine sounds. To build the real node, an ESP32 board with an ADXL345 accelerometer and a small desk fan is enough. A Raspberry Pi 5 with an AI HAT or a Jetson is optional for the vision module.
Is the course free to learn?
Yes. Every module, the notes, the worked problems, the project and the final assessment are free, and every tool used has a free route: the Edge Impulse Developer plan, Arduino IDE, Google Colab, Node-RED, Mosquitto and Wokwi. The certificate is optional.
What do I need to know first?
You should be able to read a datasheet and edit a short Arduino or Python program. Machine learning is taught from the level a tiny model needs. Machine Learning with Python and Predictive Maintenance with Machine Learning are good courses before or after this one. Node-RED for Industrial IoT goes deeper into MQTT and dashboards.
Is Edge Impulse still free now that it is part of Qualcomm?
Yes, for learning. Edge Impulse joined Qualcomm (announced March 2025) and still supports boards from other makers. Its free Developer plan gave 3 private projects, 60 minutes of compute per job and GPU training when checked in October 2026, for learning, internal R&D and pre-production; a product shipped to customers needs the paid Enterprise plan. Read the current terms before building anything for production.
Can an edge AI node trip or stop a machine?
No. The course treats every edge AI output as an advisory alarm for people, work orders and history. Trips and safety functions need safety-rated sensors and logic designed to standards such as IEC 61511 or IEC 62061. These are covered in Safety Instrumented Systems and SIL to IEC 61511.
How long does the Edge AI and TinyML for Industrial Sensing course take?
About 18 hours of video, notes and practice at your own pace, of which about 10 hours is video. The optional vibration node project takes about 18 hours more.
What certificate does the Edge AI and TinyML for Industrial Sensing course give?
An EDWartens certificate of completion. It is issued when you finish the modules and pass the 15-question final at 60 percent. It has a number anyone can verify on our site. It is not a credential from Edge Impulse, Qualcomm, Arduino, Espressif, Raspberry Pi or NVIDIA. None of them is affiliated with EDWartens.
Do I have to use the lab?
No. The lab is optional practice. You can skip it whenever you like, and it never blocks your progress, the final assessment or your certificate.
Is the lab free?
Yes. Every lab level for this course opens in your browser, free, with nothing to install.
What topics does the Edge AI and TinyML for Industrial Sensing course cover, module by module?
The Edge AI and TinyML for Industrial Sensing course has 13 modules. It opens with "Edge or cloud for plant AI: latency, bandwidth, reliability and cost", goes through "The hardware map: microcontrollers, Raspberry Pi with an AI HAT, and Jetson", "Training a tiny neural network for the LiteRT microcontroller runtime", "Quantisation, memory and latency budgets" and "Field deployment: power, enclosures, updates and drift" and ends with "Final assessment". The full list with minutes is under "What you learn" on this page.
What can I do after the Edge AI and TinyML for Industrial Sensing course?
After the Edge AI and TinyML for Industrial Sensing course: Decide which plant AI belongs on the sensor node and which in the cloud, with the data rates worked out; Choose between ESP32, Arduino Nano 33 BLE Sense, STM32, Raspberry Pi with an AI HAT and Jetson from memory, compute and power; Set a sample rate from shaft and bearing frequencies, and read vibration, motor current and sound safely. These are the practical outcomes the course is written around, and the final assessment of 15 questions checks them.
What software and hardware do I need for the Edge AI and TinyML for Industrial Sensing course, and does it cost anything?
For the Edge AI and TinyML for Industrial Sensing course, software: Edge Impulse (free Developer plan, in the browser), Arduino IDE 2, Google Colab with TensorFlow and LiteRT, Ultralytics YOLO, Mosquitto, Node-RED and the Wokwi simulator. All have a free route. Hardware: optional. Every module can be done with a smartphone as the sensor, the Wokwi simulator and public datasets. To build the real node: an ESP32 or Arduino Nano 33 BLE Sense with an ADXL345 or MPU6050 accelerometer and a small fan; a Raspberry Pi 5 with an AI HAT or a Jetson Orin Nano is optional for the vision module. The course itself is free.
Is the Edge AI and TinyML for Industrial Sensing course suitable for me, and what level is it?
Edge AI and TinyML for Industrial Sensing is rated intermediate. Intermediate. For instrumentation, maintenance, IIoT and embedded engineers, and final-year electrical, electronics and instrumentation students. You should be able to read a datasheet and edit a short Arduino or Python program. Machine Learning with Python or Predictive Maintenance with Machine Learning is a useful course before this one, but not required. It is written for instrumentation, maintenance, IIoT and embedded engineers, and final-year electrical, electronics and instrumentation students worldwide who want machine learning on the sensor node, and takes about 17.8 hours at your own pace.
Which jobs use the skills from the Edge AI and TinyML for Industrial Sensing course?
Skills from the Edge AI and TinyML for Industrial Sensing course (for example: decide which plant AI belongs on the sensor node and which in the cloud, with the data rates worked out) are used in roles such as AI engineer, machine learning engineer, data analyst and automation developer, and in embedded AI engineer work. The course gives you practical evidence of the skill; it does not promise a job, and employers decide on experience and interview.
Start free today
Edge AI and TinyML for Industrial Sensing, at your own pace. Add the certificate only if you want one.