Brains and Machines Chip Combines Analog and Digital Neurons for Sensor Data
Dec 13, 2024
Sumeet Kumar, Founder and CEO of Innatera, shares insights into their innovative spiking neural network chip tailored for sensor applications. Petrut Bogdan, Neuromorphic Architect, discusses the intricate design choices and challenges faced in creating this versatile chip. They explore the importance of energy efficiency, the integration of analog and digital neurons, and the potential for low-power applications in healthcare. Their conversation touches on the future of accessible neuromorphic engineering and the role of these technologies in enhancing smart devices.
AI Snips
Chapters
Transcript
Episode notes
Neuromorphic Computing for Sensors
- Innatera recognized the growing need for low-power AI in sensor applications.
- They saw an opportunity in the market where energy efficiency became critical.
Target Applications
- Innatera focuses on low-power, low-latency edge intelligence for sensor applications.
- They target always-on use cases like smartwatches and security cameras, where battery life is crucial.
Chip Design
- Innatera's chip includes a RISC-V CPU, various input ports, and analog and digital neurons.
- The chip's design prioritizes flexibility to cater to various sensor types and applications.