British on-device deep learning hardware platform manufacturer InferSens has launched a battery-powered ultra-low-power edge deep learning sensor for commercial and industrial applications. The hardware platform uses Syntiant’s NDP120 neural decision processor to enable highly accurate on-device processing. At the TinyML EMEA Innovation Forum 2022 in Cyprus, the company’s CTO, Jonathan Peace, will present the first legionella water system monitoring application.
Combining deep learning sensor technology with ultra-low power consumption opens a new world of edge data-intensive applications. The sensor technology will enable those edge applications which were hard or even impossible to address earlier. InferSens’ technology uses state-of-the-art silicon, edge deep learning models, and innovative mechanical and systems engineering knowledge to break the tradeoff of power consumption v/s performance.
“Ahead of the formal product launch, we are engaging with customers for pilots, under the commercial agreement, from a range of sectors, including universities, hospitals, commercial offices, hotels, and other property owners and operators,” said InferSens CEO Colin Payne.
Syntiant NDP120 neural decision processor has caught the attention of several embedded device manufacturers for edge implementation. The processor uses at memory compute and natively processes neural network instructions for enhanced data optimization, enabling InferSens to achieve on-device deep learning to process. InferSens can also deliver heavy data processing at just 1 percent of the power required by traditional processing devices.
InferSens plans to release the first product in Q1 2023, which will be a low-cost water flow and temperature sensor for monitoring and detecting Legionella risk in water systems. The device can be easily integrated into any pipe without requiring any additional cutting and plumbing. The sensor targets the commercial and public sector property market, which is approximately 7.6 million properties in the United Kingdom alone.
“We are delighted to be working with InferSens to deploy cloud-free sensor solutions with our edge AI processor technology,” said Syntiant CEO Kurt Busch. “The NDP120 delivers 25x the tensor throughput than our first-generation neural network, enabling highly accurate sensor processing with near-zero power consumption.”
InferSens CEO believes in the technology and claims that they see significant potential in multiple industries, such as multi-factor building occupancy, combined air quality, and compound sensing for buildings and smart cities where low power, data-intensive, smart sensing devices are required.