already known so again
"EDGX, BrainChip unite to create data processing units for space
Oct 10, 2023
EDGX, a Belgian SpaceTech company, and
BrainChip, a neuromorphic computing device provider, have partnered to develop data processing units for challenging environments.
The pair signed a non-binding Memorandum of Understanding at the European Space Agency’s (ESA)
EDHPC Conference in Juan-les-Pins, France.
Both companies note that space infrastructure is increasingly crucial in our daily lives as it supports essential aspects such as navigation, weather forecasting and secure communications.
Nick Destrycker, the founder and CEO of EDGX, envisions a future where interconnected satellites, fueled by advanced AI, provide real-time data connectivity for humans and machines.
“This is where BrainChip’s Akida technology comes into play, the event-based accelerator for AI at the
edge, which coincides perfectly with our vision in developing brain-inspired data processing units for space,” explains Destrycker.
To achieve self-sustainability, the space industry is witnessing a rise in satellite launches. Recognizing this demand, EDGX is currently developing a data processing unit. This unit combines AI acceleration with neuromorphic computing, enhancing next-generation satellites’ power efficiency and autonomous learning capabilities.
Laurent Hili, an engineer with expertise in microelectronics and data handling at
ESA, emphasizes Akida’s unique capability to operate in two distinct modes. The first mode is more conventional, ensuring compatibility with established neural networks. The second mode is event-based, which reduces power consumption while maintaining performance.
“Power is at a premium in the space environment,” adds Hili. “Event-based processing may easily open a door into a new realm of processing in space to do things which were not possible before.”
According to Peter van der Made, the founder and CTO at BrainChip, the Akida roadmap was designed to provide high-performance, ultra-low power processing without constant connection to the cloud, making it suitable for constrained environments like space.
“BrainChip’s focus on combining neuromorphic principles with the benefits of the deep learning ecosystem, customization and continuous learning matches EDGX’s requirements for disruptive technology. We are excited to partner with them to truly push the boundaries of what is possible,” says van der Made.
Last month, BrainChip
entered into a strategic partnership with VVDN Technologies to collaborate on the development of the Edge Box. Designed for processing data and performing computations at the network edge, this edge device harnesses the power of neuromorphic technology, drawing inspiration from the human brain. The result is a hardware platform that enhances power efficiency but also delivers outstanding performance for a wide range of edge applications, the company says."
https://www.edgeir.com/edgx-brainchip-unite-to-create-data-processing-units-for-space-20231010
and
"
BrainChip teams up with Belgium's EDGX to place its Akida neuromorphic chip in space
Published on 09-10-2023
Australian-born US company
BrainChip, which has developed a neuromorphic edge processor under the name Akida, has signed a technology agreement with Belgian start-up EDGX, founded in 2023 to develop disruptive intelligent space systems, to develop data processing units for highly demanding environments. According to the Leuven-based start-up, EDGX is developing a new generation of space computers that combine traditional artificial intelligence (AI) acceleration with neuromorphic processing to reduce the power dissipation of on-board AI, increase the adaptability of payloads and pave the way for autonomous on-board learning capabilities within next-generation satellites.
BrainChip, for its part, has developed with Akida a 100% digital event-driven AI processor that uses neuromorphic principles mimicking the human brain and analyses only the essential data detected at the point of acquisition. The idea is to process data efficiently and accurately, while limiting power consumption. In addition, the Akida chip, which can also be implemented as a block of IP in a system chip or Asic circuit, enables learning to take place locally, independently of the cloud, which, BrainChip assures us, guarantees reduced latency and improves data confidentiality and security.
"
We built the Akida technology roadmap with the objective of providing autonomous high-performance processing with very low power consumption that does not require a permanent connection to the cloud, which is essential for constrained environments such as satellites," explains Peter van der Made, founder and CTO of BrainChip.
EDGX and BrainChip formalised their collaboration by signing a non-binding memorandum of understanding at the European Space Agency (ESA) EDHPC conference held from 2 to 6 October 2023 in Juan-les-Pins (France).
"What fascinates us most about Akida technology is its ability to operate in two different modes," says
Laurent Hili, a microelectronics and data processing engineer at ESA.
On the one hand, the Akida processor can execute conventional convolutional neural networks (CNNs), making it compatible with many tried and tested CNNs. But on the other hand, in event-driven mode, it is able to reduce power demand by several orders of magnitude. Bearing in mind that consumption is a key concept in the space environment, this type of operation could open the door to a new area of processing in space to achieve things that weren't possible before." 
It should be noted that, in early October, BrainChip announced the early availability of the second-generation Akida platform, which can be used in applications as diverse as the connected home, smart city, industrial and automotive, and is designed for extremely eco-efficient processing of complex neural network models on edge devices.
According to BrainChip, support for 8-bit weight values and activation levels, as well as long-range skip connections, extends the reach of fully hardware-accelerated models on the Akida platform. The second generation of the Akida platform now incorporates TENN (Temporal Event-based Neural Networks) spatiotemporal convolutions, a new class of neural networks that are distinguished by their lightness and efficiency in the use of resources, and which are particularly well suited to the needs of applications requiring continuous processing of raw data containing significant temporal information. Examples include video analysis, target tracking, audio classification, MRI and CT scan analysis for vital sign prediction, and time series analysis for predictive maintenance.
This technology, combined with the hardware acceleration of Vision Transformers (ViT) models, paves the way for edge devices capable of processing advanced vision and video applications consuming, according to the company, only a few milliwatts at sensor level (a few microwatts for audio or similar applications).
"
Processing generative AI and large LLM language models at the edge is essential for intelligent situational awareness in verticals ranging from manufacturing to healthcare to defence," says Jean-Luc Chatelain
**, managing director of Verax Capital Advisors.
Disruptive innovations such as BrainChip's TENN technology supporting Vision Tranformers based on neuromorphic principles can provide compelling solutions in ultra-low power, small form factor devices without compromising accuracy."
You can also follow our news on L'Embarqué's LinkedIN showcase dedicated to artificial intelligence in embedded systems:
Embedded-IA
You can also follow our news on L'Embarqué's LinkedIN showcase dedicated to the aerospace and defence market:
Embedded-Aerospace
https://www.lembarque.com/article/b...cer-sa-puce-neuromorphique-akida-dans-lespace
**Episode 23 – BrainChip Talks AI Innovation with Accenture’s Managing Director and Global Chief Technology Officer of Accenture Applied Intelligence Jean-Luc Chatelain
That’s enormous. I also like the the sound of 100 million expected devices 
