In another AI-focussed interview with ipXchange at CES 2024, Guy chats with David, CEO of AIStorm, about another innovative AI solution that utilises the low power consumption of analogue circuit design.
In the case of AIStorm, a charge-domain-based computing system runs on each pixel of a specially designed image sensor, enabling high performance that would require a multi-GHz digital processor to do the same task at the same speed. This enables better, faster image contextualisation and target recognition for industrial cameras and sensors, with a power consumption that is typically 200x smaller than a digital solution.
As David explains, a lot of industries want AI-augmented cameras – retail, smart building, industrial, etc. – but standard retrofitted cameras just don’t last long on battery power. AIStorm’s solution can last for 1-2 years on standard AA batteries and can network with each other to provide better performance for AI workloads like people tracking, traffic pattern recognition, facial recognition, and eye- and interest-tracking.
A particularly interesting example that David gives is ‘trusted customer’ functionality for smart retail, which allows certain customers access to cabinets that are usually locked, for instance in a pharmaceutical setting.
AIStorm’s technology brings a new viability to low-maintenance AIoT installations that use sound and image classification to better understand and interact with users. AIStorm provides a complete, compact system on chip, with everything you’ll need (image detection, processing, software) except for the battery and the lens.
Learn about ‘Mantis’ by following the link to the board page below, where you can apply to evaluate this technology for use in a commercial project. We’ve also included a brief overview of ‘Cheetah’, which takes AIStorm’s high-speed imaging technology and puts it into a dedicated chip for up to 10 kFPS operation!