The group is led by Prof. Massimo Alioto, and involves amazing people who are passionate about integrated circuit technology and its impact on the life of (many) people. Energy efficiency is the focal point of our activity, as its improvement can bring large benefits to our environment and drive the evolution of everyday electronic objects through their form factor, battery life and enhanced capabilities (e.g., mobile systems, wearables, smart objects, Internet of Things).
Below there is a short description of our interests, location, vision and mission. Please, contact Prof. Massimo Alioto for further information. Our press releases and the impact of our research on worldwide media are summarized in the Press page. We also like sharing our vision and sense of perspective over a number of public events and publications, as summarized under “Sharing thoughts, publicly“).
Feel free to explore our short videos and experimental demonstrations on several new technologies that are being developed in our group: our research and its impact on the world in a nutshell. Some of our chips are showcased in the chip gallery.
Hope you will enjoy!
VALUE PROPOSITION : we enable AI vision at scale, anywhere, no strings (wires) attached.
THE TECHNOLOGY: AI silicon chip for tiny cameras with zero wires, 10-year battery, always-on, highly-capable AI. This is made possible by its lowest power consumption in the world (<1 mW at 30 fps, >30X more power efficient than any research prototype, 100X over commercial).
End-to-end innovative system from advanced AI image sensor, to intermediate activity filtering, to AI accelerator and wireless connectivity. Each with best-in-class energy efficiency.
AI is completely software-programmable for self-reliant vision and frictionless ingestion of AI models created by the user, or downloaded from public repositories.
Privacy-preserving and highly-scalable since processing is performed locally. Rapid and effortless installation (no wires), maintenance-free (long battery life), also retrofit in areas w/o infrastructure.
Example of captured information: novel content is detected, performing AI recognition (face recognition shown here).
Example with no novel content: compute-intensive AI is suppressed to keep average power very low.
THE SCIENCE: CogniVision is the outcome of a successful 5-year NRF-sponsored CRP program, and the research innovation of 15 world-class researchers.
Some publications defining today’s state of the art (out of 60+ and 8 patents):
A. Gupta, S. Kumar, V. Konandur, S. Taneja, M. Alioto, “Visual Content-Agnostic Novelty Detection Engine with 2.4 pJ/pixel Energy and Two-Order (…),” VLSI Symp. 2023
J. Vohra, A. Gupta, M. Alioto, “Imager with In-Sensor Event Detection and Morphological Transformations with 2.9 pJ/pixel×frame Object Segmentation (…),” ISSCC 2024
A. Gupta, J. Vohra, M. Alioto, “CogniVision: End-to-End SoC for Always-on Smart Vision with mW Power in 40nm,” IEEE VLSI Symp. 2024
DEVELOPMENT STAGE: moving from silicon prototype to product, looking for agencies and industry partners to test the technology in field. Deployment with zero wires demonstrate unprecedented ubiquitous vision for surveillance, monitoring, area access control, operational safety, under-served areas, temporary vision deployments. Deployed AI cameras will add capabilities to existing cloud, resiliently and efficiently performing data analytics in place (distributed AI).
At a scale, cost and in places where this was not possible before.
VISIT US AT
