In the last of ipXchange’s pre-show ipXperiences with the partners who will be on our Booth (4A-101) at Embedded World 2024, Eamon chats with GP Singh, CEO of Ambient Scientific, for a deeper understanding of exactly how this disruptive company has managed to enable AI processing at 60 µW or less in a chip so economically priced that an AI-enabled end product can now be sold for around $20!
To start, Eamon asks GP about the inner workings of Ambient Scientific’s GPX-5 and GPX-10 ultra-low-power always-on AI processors, focussing on Ambient Scientific’s unique take on an analogue matrix computing architecture and how this enables ultra-low-power deep-learning artificial intelligence that can be run using the smallest batteries and even harvested ambient energy. Discussion then moves onto the integrated Cortex-M core and where this fits into the workload while helping designers to create products that can fully utilise AI at the edge, using development ecosystems that they will already be familiar with.
But the main topic of this interview is the applications that Ambient Scientific’s chips enable at a new price point that seems impossible for the amount of on-device intelligence. GP shows us several reference designs ranging from compact event-detection lockets for health and safety monitoring to ultra-compact face and weapon recognition camera systems that can be installed in great numbers and easily replaced thanks to the ultra-low cost.
Enjoy this in-depth discussion, and if it’s got you thinking about implementing AI, head to the board page using the link below to apply to evaluate this technology today. And don’t forget that you can see it in person at Embedded World 2024, booth 4A-101. This year is going to be ipXchange’s best yet!
Keep designing!
Love listening to GP talk about AI? Check out our first interview with him from CES 2024:
Want ‘Michelin-Star’ AI? Master the basic ingredients first! Analogue matrix computing at 60 µW
You must be a member to leave a comment. Sign up or Sign in