I recently wrote a blog post to outline a modular framework I am creating to accelerate development of compute hardware and lower product risk and cost at low/mid range volume. Happy to answer any questions here.
24 karma · joined March 8, 2012
Also, the software running on open source hardware devices is often an afterthought and there are very few devices that have hardware and software co-designed and tightly integrated to work together seamlessly and be developer friendly.
This drove me to design Ubo Pod that closes the gap between a developer-friendly platform and a polished product ready for end-user deployment.
Ubo creates a multi-modal user experience by bringing together voice, vision, and declarative GUI (embedded and WebUI). It supports over 50 cloud AI service providers (bring your API keys) as well as on-device/on-premise model hosting options.
The work behind Ubo Pod is the culmination of my decade long experience designing and building hardware products from initial concept to mass production. I am proud of the work me and my teammates have put into designing polished hardware and sub-modules, sophisticated software stack, agile manufacturing, global supply chain, automated test/QA.
I hope the open source hardware and software design can serve as a blueprint and inspiration for a new class of open devices that can tailor to both everyday consumers and professional developers.
In this video, I am showing testing of a reference design and overall plan to implement this.
The IC that does the conversion is from Lontium Semiconductor (LT6711A).
If you like this idea and would like to see it come to life, please let me know if the comments and let me know if you have any thoughts on improving the design or other features you'd like to see.
I am using Ubo pod hardware (own product) in this demo but you don't have to. You can just use a bare Raspberry Pi and use the emulated GUI inside the web UI (instead of the built-in GUI). All you need to do is to flash MicroSD card with image available on Github page ubopod/ubo_app/releases.
Do you use Twingate in your homelab? If so, please comment on this post and share you thought with me.
The hardware and software is in a great shape and I have already manufactured a small batch.
I would like to get some community feedback before launching on Kickstarter to better understand what to improve and potentially add to the upcoming campaign.
Looking forward to hearing your thoughts.
Ubo app:
- offers a GUI (shown inside browser or hardware) to control your Raspberry Pi and Linux utilities. - helps with setting up WiFi even if your credentials change after initial setup. - gives your Pi virtual speakers, (and soon) microphones, and camera. That means, you can play audio on Pi without any audio hardware and hear the audio out of your phone/laptop speakers (virtual speakers). Same for mics and camera. lets you set up remote access (RPi Connect and VSCode Tunnel) using the GUI and configure different system/app settings, install Dockerized applications, etc. - easy configuration of LLM engines and voice assistant - offers a language agnostic gRPC API that you can communicate with from any programming language
-Here's a quick example of installing a Dockerized app (e.g. Jupyter Notebook) in a few seconds:
In its current form, it offers a set of tools (development toolkit) for embedded system application developers to perform deep hardware integrations with ease. The hardware system is designed with fine-grain modularity and powerful UX which can be easily customized for enterprise clients with minimal to no tooling fees.
In the near future, we’d be opening up ai-assisted workflows and agents to help developers build systems much faster. We are building an agentic system that chats with developers starting with product description and works with them to derive requirements, user stories, acceptance criteria, test scenarios/cases down to test code and application code that runs on hardware for validation.
The current hardware manifestation of the first-iteration concept, is a hardware pod that developers can use to build in the following application spaces:
- Smart home automation - Personal data storage and hosting - Edge AI & ML applications - IoT sensing & monitoring - Self-hosted task runners - Voice assistant applications - Networking applications (hotspot, remote gateways, etc.)
The product is constantly evolving and it will be introduced into the market in steps, initially starting with the Ubo Pod dev kit and later the Ubo factory AI-assisted development tools that can be used to create infinite variations of the base concept. You can think of Ubo Pod as the chassis/framework for building embedded systems.