Would you take a ride in an autonomous vehicle? And what happens if a pedestrian suddenly steps out on the road? Would you trust the vehicle to make that split-second decision that a human would take to save a life or prevent even the most mundane of accidents?
Responding to unexpected situations requires ‘ultra-fast response’ technologies. These depend on communication and processing ‘in the cloud’, which can sometimes take too long due to latency. Local, ‘edge computing’ can speed things up. However, this means installing large and energy-hungry computer systems into vehicles. Of course, this isn’t just a problem for vehicles. Ultra-fast responses are essential in many industries and even in your mobile phone.
A new European Commission-funded Pathfinder project HYBRAIN (Hybrid electronic-photonic architectures for brain-inspired computing) aims to overcome these barriers by delivering a new computing system that is inspired by the human brain. Coordinated by Professor Wilfred van der Wiel, University of Twente, the project gathers key partners from Oxford, Münster, Pisa and Zürich. The project will develop an “HYBRAIN system” that is both super-fast, consumes very little energy, and is headed to a make real impact on ‘ultra-fast response’ technologies.
Cloud computing has been considered the best solution to keep data and computer processing at a distance. But it is now becoming increasingly important to move it close to the actual ‘operation’ and start working locally again; this is what is also known as ‘edge computing’. This avoids the ‘latency’, which despite the upcoming and fast mobile standards like 5G and 6G, can still be too long. This does present a dilemma, though. Moving heavy computing power to a local application is not ideal as a classic computing approach implies a lot of data traffic between the processor and memory. This is not how our brain works, where memory and processing are part of the same process. And in the new HYBRAIN project, researchers will combine a number of highly innovative solutions, based exactly on how our brain works. These solutions include ‘in memory computing’ and an evolutionary system that is disordered by itself but can nevertheless detect complex patterns.
It all starts by gathering huge amounts of data. This, for example, can be visual data. Integrated photonics is very good at this. That is why the input of the artificial intelligence system consists of a photonic processor, working with light. The data will be divided among two learning networks: one is a network based on in-memory computing, consisting of so-called ‘memristors’: resistors that can memorize their settings even when switched off. A network like this is capable of performing linear operations, like multiplying and adding large streams of data.
Apart from that, a disordered network built of nanomaterials, developed at the University of Twente, has proven to be able to deal with complex, non-linear operations. These two types of networks will first get a learning phase (inference), after which they can make choices by themselves. Of course, this approach implies connections between systems, but there is no large-scale data transport in two directions, as in traditional computers.
We could even ask Can’t the entire system be built in photonics? The issue here would be the size: photonic components and circuits are relatively small. However, no conversion to electric signals would indeed be needed. “Still, it will be very exciting indeed”, Wilfred van der Wiel says. “We connect three types of technology each in a very experimental stage. They don’t have the ‘history’ and huge industrial power that the current CMOS chips have”. Still, if the project results in a new system that succeeds in reaching an energy consumption that comes even close to that of the human brain – a few hundred Watts – a major step will have been made.
Aside from autonomous vehicles, edge computing can be necessary for particle accelerators such as in CERN in Geneva and other ‘ultrafast response’ technologies and even some of the apps on our smartphones use local computing power, instead of cloud computing.
The HYBRAIN project is led by the Center for Brain-Inspired Nano Systems (BRAINS) of the University of Twente’s MESA+ Institute. Supporting the coordinator in this ambitious endeavour is a lean consortium with specific skills from the University of Münster (WWU, Germany), the University of Oxford (UK), the Italian SME, Trust-IT, and IBM Research in Zürich (CH). The University of Twente (UT) has a clear focus on sustainable hardware and strategic partnerships and HYBRAIN perfectly fits perfectly into its core strategy and mission.
HYBRAIN is the first University of Twente project to be awarded within the new Horizon Europe programme with a budget of 3 million euros and a timeframe of four years. The kick-off is taking place for two days, between 23-24 May 2022, at the University of Twente in Enschede, Netherlands.
For more information contact us at email@example.com
Trust-IT is leading the HYBRAIN's Dissemination and Exploitation activities. As such TRUST-IT will define a clear communication strategy by defining the objectives, Key Exploitable Results (KERs), impacts, and target stakeholders of the project. This will allow us to identify the right channels to use and tailor our communication activities for each stakeholder and maximise the community engagement by sharing the information that matters the most to them.
We will also develop, host, and maintain a UX and SEO-optimised public website that will be the main platform for the HYBRAIN community to gain detailed information and insights on the project's activities and results.
In preparation for the today’s kick-off, 23 May, we have already created the project:
Do you need a communication & dissemination partner to support your research & science results in your EU-funded project? Get in touch with us!