Microelectronics made its appearance in the 1960s with the arrival of semiconductors and the transistor. The CEA (Commissariat à l’énergie atomique et aux énergies alternatives) was one of the pioneers of this industrial sector, notably with the creation of CEA-Leti (Laboratoire d’électronique et de technologie de l’information) in 1967. At the Leti Innovation Days in 2021, CEA-Leti presented its µWAI (micro WAY) imager, on which it is continuing to work with STMicroelectronics.

Microelectronics is at the heart of the devices we use every day: phones, computers, cars and connected objects.

The µWAI stand-alone imager

The size of a 1 € coin, µWAI features a co-designed reading and processing architecture with an optimized algorithmic pipeline, facial recognition results from a sequence of elementary algorithms that provides very low power consumption and the use of compact silicon to reduce costs.

According to CEA-Leti, this is the first smart image sensor combining :

• Automatic exposure for all lighting conditions and 88 dB dynamic range.
• Motion detection and feature extraction for event-based operation.
• AI-based object recognition for highly reliable identification.

These key features enable highly reliable decision making for a few tens of pJ/pixel/image (pJ/pixel/frame measures the energy expended by each pixel for each unique image in an image frame). This imager would be much more efficient than currently marketed systems, a typical implementation requires about 10,000 times more energy than µWAI.

Its power consumption is 3-6 µW, which is required for IoT, the lifetime of the always-on CR1025 button cell is estimated to be five years.

Applications of µWAI

The imager’s applications and functions include:

• Mobile devices: automatic switching, face identification;
• Smart home: contactless smart switches for home, sports and entertainment devices;
• Smart building/smart home: facial recognition, people counting, alarm triggering;
• Automotive: in-cabin situational awareness, driver identification;
• Automotive: situational awareness outside the cabin, intelligent door unlocking system.

Looking for new applications

The CEA-Leti team turned to STMelectronics to continue its research, a company with which it often cooperates. In 1972, the institute created a subsidiary to promote its work in the field of semiconductors: Efcis (Études et fabrication de circuits intégrés spéciaux), which was renamed SGS-Thomson after Thomson-Brandt’s investments and is now STMelectronics.

Antoine Dupret, head of industrial partnerships at CEA LETI, states:

“The recognition engine is optimized to recognize faces when motion is detected. The CEA-Leti team is working hand-in-hand with STMicroelectronics to develop specific smart imaging products as we look to extend the technology to other use cases.”

He adds:

“We aim to adapt the recognition engine as an embedded IP in various cameras and optimize the imager’s performance to our partner’s customer requirements.”

Translated from Microélectronique : Focus sur µWAI, l’imageur autonome du CEA-Leti