Buzzword Bingo (or: which modern technologies make sense for embedded?)

The Pragmatic Embedded Podcast · 13 Nov 2024 · 49m

AI in practice: 4 real uses of AI described

What this episode covers

Hosts Jeff Gable and Luca Ingianni evaluate modern development buzzwords for embedded systems, discussing the applicability of web technologies like Kubernetes and Docker, the nuances of CI/CD pipelines, and the practical adoption of Rust and modern C++. They highlight specific tools such as Memfault, Mender, and Renode for observability and simulation, and analyze the role of AI in both device-side machine learning and developer productivity.

AI in practice

Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.

Engineering & simulationChatGPT

The speaker uses ChatGPT to generate code skeletons and iteratively refine functions, significantly speeding up the coding process compared to writing from scratch.

“you can say, give me a skeleton for this, and then you can get into a really nice iterative loop where you say, okay, that was nice, but do it more in this style, or change this function to have, accept function pointers for this. And you can very quickly bang out code that would have taken you quite a lot of writing…”

Jeff Gable · Hear it at 44:25

Engineering & simulation

The speaker used an LLM to convert tests from one framework to another and received recommendations for refactoring the code into a data-driven style.

“I asked, and L&M, not even ChatGPT, something else, to convert tests from one testing framework to another testing framework for me. And not only did it do that, but it also recommended refractoring, it said, you should really do this in a data driven style.”

Luca Ingianni · Hear it at 45:31

Engineering & simulationChatGPT

The speaker uses ChatGPT to learn C++ features by incrementally converting C code to different C++ standards through iterative prompts.

“you can say, okay, I've got this. I've got this C code, convert it to C ++11. And just, you don't use it blindly, but you use it to learn, okay, what are the features in C ++11 that, that, you know, and you can, again, go back and forth with it…”

Jeff Gable · Hear it at 46:40

Engineering & simulationMicrosoft Copilot

The speaker mentions Microsoft Copilot as a significant AI coding tool used by developers, though he admits he hasn't used it extensively himself.

“Microsoft Copilot is a big one. I haven't, I'm not the expert. I haven't used all of these tools. I just used, basically, chat GPT and clawed, but again, listeners, this is one where we really want to hear from listeners as to what tools you've used…”

Jeff Gable · Hear it at 46:40

Mentioned

Companies
SEGGER, Mathworks, STMicroelectronics
Tools
Memfault, Mender, Renode, Simulink, Docker, Kubernetes, Yocto, Buildroot, STM Viewer, Bitbucket, GitHub, GitLab, ChatGPT
Also
Rust, C++, Jacob Beningo, Nile Cooling

Other shows on the same ground, closest first.

  • Embedded.fm · 51m

    266: Drive off the End of the Universe

    Hosts Elecia White and Christopher White discuss the utility of emulation and simulation tools like QEMU and Renode for embedded development, emphasizing their value in continuous integration and early software testing. They address listener questions regarding the selection and maintenance of third-party libraries, the value of professional training versus conferences, and career advice for embedded engineers interested in machine learning. The episode concludes with a brainstorming session on future 'basics' topics, including pre-RTOS control flow, design patterns, and RTOS fundamentals.

  • Coredump Sessions · 55m

    #007: AI, Open Source, and the Future of Embedded Development: How Much Code Will We Actually Write?

    François Baldassari, CEO of Memfault, and Thomas Sarlandie, Field CTO of Memfault, host a panel with Jonathan Beri, CEO of Golioth, and Dan Mangum, CTO of Golioth, to discuss the intersection of AI, open source, and embedded development. The group analyzes the maturity of edge AI for audio and video processing, the challenges of integrating LLMs into hardware due to rapid model iteration, and the current state of using AI for firmware development. They conclude that while AI is transforming code generation and review, open source remains critical for training models and providing trusted, complex infrastructure like Zephyr and FreeRTOS.

  • Embedded.fm · 1h 05m

    400: A Really Long Time

    Hosts Elecia White and Christopher White reflect on nine years of the Embedded.fm podcast, discussing the evolution of embedded systems from 2013 to 2022. They highlight improvements in tools like VS Code and open-source RTOSs, while criticizing persistent issues with vendor datasheets and the lack of standardized device drivers. The conversation also covers the shift in hardware capabilities, the rise of TinyML, and the hosts' personal experiences with consulting, Fitbit, and community building.

  • Embedded.fm · 1h 19m

    430: Broken Toys All Around Me

    Hosts Elecia White and Chris White discuss a wide range of embedded systems topics, including sourcing components from distributors like DigiKey and Seeed Studio, and using tools such as Visual Studio Code with the Cortex-Debug extension for STM32 development. They analyze the engineering challenges of accelerometer-based crash detection in iPhone 14 and Apple Watch, referencing a past Fitbit rollercoaster bug, and debate the industry's transition from C to Rust. The conversation also covers career advice for embedded engineers, the use of AI tools like DALL-E for content, and the importance of systems engineering and documentation.

  • Embedded.fm · 56m

    256: Agglomeration

    Chris White and Elecia White discuss the ST Micro F1 series, comparing the use of HAL, Cube, and bare metal programming for embedded systems. They review the Hippocratic Oath for Connected Medical Devices from IAmTheCalvary.org and Make Magazine's tips for securing DIY IoT projects. The hosts also explore the Rockstar Programming Language specification, C++ expressiveness, and Elecia's recent work with Python and JSON.

The summary, chapters and topics are generated from the episode’s audio by this directory, not written by The Pragmatic Embedded Podcast. The episode itself belongs to its publisher.