Impact

Rising to the Grand Challenges

ICNS research addresses three connected, critical problems: How can we (i) continue to enhance computing power beyond the end of Moore’s Law to (ii) process and analyse the exponentially growing amounts of data that support modern society while (iii) conserving energy? Learn more about these problems on our Research page.

We study and mimic biological systems and sensory organs to develop sensors, processors and algorithms that out-perform conventional digital solutions. By increasing computational  power and efficiency, ICNS researchers will help to solve many of the Grand Challenges for engineering in the 21st century (opens in a new window) including: securing space infrastructure, enhancing health informatics, securing cyberspace, improving the utility of  urban infrastructure, advancing personalised learning, enhancing virtual reality, and engineering the next generation of tools for scientific discovery.

We’re addressing the Grand Challenge to reverse engineer the brain (opens in a new window). Using architecture, processing, and coding inspired by brains and nervous systems and applying world-leading models of neural network function, we can create artificial elements of brains with processing power far greater than that of current computers, as well as developing solutions for neurological damage or disease. Our fast, low-energy, compact and robust neuromorphic systems are especially ideal for distributed computing, mobile devices and autonomous applications in challenging and remote environments.

Globally, governments and corporations are investing millions in neuromorphic R&D because they understand its huge potential for positive impact.

Discover more specific examples of the real-world impact of our research by clicking on the headings below.

Space Situational Analysis

Our social, economic, scientific and defence systems depend on satellite communications and Earth observations from space. In Earth’s orbit, there are now around 5,000 satellites and 34,000 pieces of junk larger than 10cm (discarded parts from spacecraft or satellites and debris from explosions and collisions). Space Situational Analysis (SSA) monitors these orbiting objects to predict collisions and descents that could damage essential infrastructure in space or on the ground, as well as risking lives.

Most digital cameras employ a charge-coupled device (CCD) to convert light into electrons. When applied to SSA, CCD sensors collect high-resolution images continuously, mostly of empty space, generating a huge amount of data that must be transmitted and processed, though most of it will be ultimately discarded. This wastes time and energy. An additional problem is that CCD sensors are overwhelmed by daylight which also results in the need to deploy a network of SSA systems across large geographic areas chasing the night time sky. This is inefficient and costly.

Astrosite

Astrosite

To solve these problems, ICNS researchers are developing unique, highly efficient neuromorphic sensors that can track and predict the trajectory of orbiting objects in real-time, even in bright daylight, enabling 24/7 observation. Our camera is an asynchronous array of artificial neurons (one per pixel). Each neuron operates independently with microsecond response times, only reporting when it detects a change in light intensity, keeping data sparse and using minimal energy. The camera doesn't need to be stationary to operate, so it can be mounted onto a vehicle or even its own satellite. We have successfully deployed a neuromorphic camera in a mobile ground observatory (the Astrosite) for SSA. This world-leading approach is building game-changing SSA capability for Australia. Stay updated on this project by searching for #astrosite. For more information, contact Associate Professor Greg Cohen.

Partners

Related video

Autonomous drones

Event-based neuromorphic sensors allow autonomous drones to analyse their surroundings quickly and accurately for navigation and collision avoidance. Our sensors use far less energy than conventional cameras and visual processors, so their batteries can be smaller, making drones lighter and more manoeuvrable. Our sensors work well in dim or bright environments and since they can detect an impending collision with microsecond precision, the drone can avoid it even when flying at high speed, at night-time or in dazzling sunlight.

Related videos