Innatera Nanosystems BV, a leading developer of neuromorphic processors that roughly mimic the functioning of the human brain, today announced the launch of Pulsar, its first commercially available mass-market microcontroller for edge devices.

Pulsar uses a specialized architecture based on Spiking Neural Networks, which differs from traditional artificial intelligence hardware by processing data in a way that mirrors the brain’s focus on changes in input. This enables the processor to operate at extremely low power and latency, making it ideal for real-time signal processing and decision-making.

In an interview with SiliconANGLE, Innatera co-founder and Chief Executive Sumeet Kumar said Pulsar goes even further and combines its neuromorphic edge compute capabilities with a high-performance RISC-V processor and dedicated AI accelerators for additional versatility.

“Pulsar consumes 500 times less energy and processes data 100 times faster than conventional microprocessor approaches,” Kumar said. “We can extend the battery life of a traditional video doorbell from a few weeks between recharges to one and a half years simply by being more efficient about how we process data.”

According to Kumar, Pulsar works best by helping filter and interpret sensor data locally and thus keeps the main processor asleep until it’s truly needed, in some cases this eliminates the need for a main application processor or cloud computing, thus extending battery life by orders of magnitude.

The new chip makes always-on intelligence for sensors viable, including for sub-millisecond gesture recognition for wearables, energy-efficient object detection in smart homes, the capability to detect and count humans and more.

“Innatera’s Pulsar chip has the potential to redefine what’s possible at the edge,” says David Harold, senior analyst at Jon Peddie Research. “By using brain-inspired SNNs, it brings real-time processing to ultra-low-power devices without leaning on the cloud. That means sensors that can think for themselves — faster responses, lower energy use, and smarter performance across everything from wearables to industrial systems.”

The new chip makes always-on intelligence for sensors viable — including sub-millisecond gesture recognition for wearables, energy-efficient object detection in smart homes, the capability to detect and count humans and more.

Kumar said Innatera is already working with companies such as system on a chip maker Socionext Inc. in the development of products using Pulsar’s neuromorphic processing. In one use case, he showed off how it could be used as part of a “human” detection system for a door that reacts only when a person approaches.

“The combination of Innatera’s Spiking Neural Processor and Socionext’s highly integrated, sophisticated radar sensor technology introduces a powerful new approach to reducing power consumption and minimizing false detections in challenging applications, such as battery-powered devices,” said Matthias Neumann, senior marketing manager smart sensor and smart devices at Socionext.

To support the ecosystem of devices that will use Pulsar, Innatera is launching a developer program, now open to early adopters. The company includes a software development kit called Talamo, which makes developing for its neuromorphic processor approachable. According to Kumar, developers can build their own spiking models from scratch in a PyTorch-based environment, then simulate, optimize and deploy them.

“Processing at the edge isn’t going to be optional anymore — it’s going to become inevitable,” Kumar said.

Image: SiliconANGLE/Microsoft Designer