Feature: Embedded design
Figure 1: Comparison between Cortex-M7, M55 and M85 for traditional and ML workloads
Rethinking the embedded development workflow Traditional embedded development typically revolves around an IDE, a compiler and a hardware debugger. While that workflow still applies, ML projects introduce additional complexity; see Figure 3. In practice, much of the work occurs outside the embedded IDE. A typical ML-enabled embedded project includes a number of distinct stages: 1. Data acquisition 2. Data cleaning and analysis 3. Model design 4. Training and optimisation 5. Deployment to embedded hardware 6. Validation and testing. Each stage oſten uses different tools and environments. For
many embedded engineers, this workflow may initially feel unfamiliar. However, the growing ecosystem around embedded ML is making the process increasingly accessible.
Development environments for embedded ML Because training ML models requires substantial computational power, most engineers rely on powerful workstations or cloud- based computing resources. Fortunately, modern development environments make this remarkably easy.
Jupyter Notebooks and Jupyter Labs A widely used environment for scientific computing is the Jupyter Notebook. Jupyter notebooks combine two key elements:
• Text cells that explain concepts or describe the workflow; • Code cells that execute IPython (Interactive Python) scripts. Tis structure makes notebooks ideal for experimenting with
algorithms and documenting the design process simultaneously. Each code cell can be executed individually, allowing engineers to progress through the training and evaluation of an ML model interactively.
Figure 2: Arm MCU and NPU offerings
www.electronicsworld.co.uk May 2026 17
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44
