We're building a Movidius X Carrier Board for Raspberry Pi Compute Module, increasing the Pi's object detection by 30x and dropping its CPU load to 0%, making realtime embedded machine learning available to all! Click here to participate in the design which is in progress, or join our email list to get updates and access to early-bird prices.
Play Video

Easy to Use

Aenean condimentum accumsan dolor, a imperdiet lectus tempor in. Aenean gravida laoreet exterior.

design for free

Nullam rhoncus fermentum dignissim. Nunc tortor ex, interdum a sem sit amet, cursus pulvinar muliar.

Made with Love

Curabitur finibus ligula sit amet pellentesque placerat. Aliquam semper scelerisque turpis equalo.

We're Building a Movidius X Carrier Board for Raspberry Pi Compute Module, which means...

Embedded machine learning for all

The AiPi enables for embedded machine learning what the original Pi did for software programming education. It allows anyone to get up and running with a complete solution simple and fast. And the key difference between AiPi and other solutions (e.g. using a NCS2 with the Pi) is that crossing the threshold to get to real-time enables all sort of applications which are not possible when you’re running at max CPU on the Pi, and still only achieving 3-4 FPS in your end application. With this carrier board you get real-time depth and real-time neural inference results all while leaving the Pi CPU at 0%. Which is a far cry from the existing solutions, where just to get a 3-4FPS solution the Pi is completely maxed out, and you have no room left to write your code on top of these ‘neural network tools’, no room to actually implement what you need to do in your industry.

Increase in Object Detection FPS compared to Raspberry Pi 3B+ itself
Increase in Object Detection FPS compared to Raspberry Pi 3B+ with NCS2
increase in stereo depth compared to Raspberry Pi (Assuming Pi 3B+ can do 20FPS)
Reduction in Raspberry Pi CPU load (from ~100% to ~0%)
How does it work?

This is what AiPI Sees.

We’re working to make a board which leverages the Myriad X to do the depth calculation (and de-warp/etc) directly while also doing the neural network side (the object detection). This should take the whole system from ~3FPS to ~30FPS, while reducing cost. Here’s a demo with an Intel RealSense D435 + Raspberry Pi 3B + NCS1.
Saving lives with tech.

Here's a working Firmware/Software Prototype!

We’re building a product to save cyclists’ lives. Here’s a working prototype!

More Brains, Please.

Meet the Core Team

AiPi is a collaborative effort. EVERYBODY is welcome to contribute (yep, that includes you) regardless of experience, and here is the perfect place to start. Below are the folks who are hammering away at this with massive chunks of their time.

Brandon is an EE who’s passionate about building things that have a positive impact on the world. He LOVED leading the UniFi team for this reason because it connected businesses (and individuals) to the power of the internet reliably and efficiently - at a fraction of the cost of incumbent competitors - and typically with much higher performance and stability. He’s hoping to leverage the experience from UniFi to be help the world in an even more direct way - by saving bike commuters' lives - and thereby bringing the harmonious biking atmosphere of the Netherlands to other parts of the world.
Brandon Gilles
Brian is an EE hardware design engineer with experience from the aerospace and data storage industries. He has an insatiable and eclectic curiosity which allows him to learn quickly and solve interdisciplinary problems in a way far beyond the scope of a traditional electrical engineer. Brian’s passionate about leaving the world a better place than he found it, so the opportunity to to build AiPi and Commute Guardian is a natural fit.
Brian Weinstein
Patrick is a Software Developer who has also worked as a UX Designer, UI Designer, and Copywriter. He is passionate about presenting complicated (and often boring) data in a relatable way so that it becomes easy and FUN for "regular" people to consume. He was hit by a motorist while commuting by bike in 2013, and is therefore extra motivated to create public awareness for such a simple (to consumers), life-saving product. Which means, because AiPi is a required part of the Commute Guardian journey, that he's motivated to bring AiPi to developers, creating the path for many other peoples' projects to flourish along the way.
Patrick Griffith
We'd LOVE to have you

Become involved

AiPi is a collaborative effort. Want to help? Want to give input on what the design should be, or other designs you’d want instead? Or simply want to see more options for embedded machine learning? Check out our community.